4

DOCUMENT RESUME

ED 217

(AUTHORTITLE

'INSTITUTIONSPONS AGENCYPUB DATEGRANTNOTE

EDRS PRICEDESCRIPTORS

.

CE 032 933

Taylor, Patricia A:; Grandjean, Burke D.Schooling; Trainirig,.and,Patterns of OccupationalChange 'among Career Civil Servapts, 1963-1977. FinalReport.

(ED), Washington,\DC.Sweet Briar Coll., Va.National Inst. of EducationJun 81

'NIE-G-78-0005

Mi'01/PC05 PlUs Postage.AdUlts; *Caree Change; Career Laddeils; *EducationWork Relitionship;,*FederalGovernment; *GoverwentEmployees; Midlife T sitions; *OccupationalMobility; *Transfer Of Training; VocationalAdjustment

ABSTRACT'This project examined the careers of a one percent,

sample of federal civil servants for the years 1963 through 1977. Theareas of emplOyment investigated included economic returns toschooling and experience/ factors affecting promotion and turnover,and occupational career la dens. A series of hypotheses andobjectives specified at thd outset wee subjected'to multivariateestatisticaianalysis using he official personnel, records of morethan 68,000 federal.employeeS. The results show that the mainconclusions of human capifl research in economics and statusattainmet research in sociologyholddor the federal government.However, internal labor market chAfacteristics must also be--considered for a full understanding of civil service careers. The:findings suggest 'that mobi'ity between major occupational categoriesis infrequent in the federal service, and that_ when it does occur theattribUtes of origin and destination jobs are not similar. Attributecontinuity is greatest for mobility involving a change of grade ofdetailed occupation without a change of Major categories, for .1

ipploye- in mid career, .not new entrants; and for two_ of the nineattributes considered -- knowledge required-by the job and its physical

%demands. The implication for educational practice is that a concernYtor teaching transferable skills to prepare students for a career ofcfccupational change should not sacrifice thorough mastery df specificsubject matters. (Author/KC)

, ***********************************'*****;***k*************************Reproductions supplied by EDRS are tHe best that call be made

*, from the original document.*********************************************,**************************

/1-

40. .1

A

C

4

$

SCHOOLING, TRAINING, AND PATTEAS_OF OCCUPATIONAL'CHANGE AMONG'CAREER CIVIL SERVANTS, 1963-1977*

Final Report to the National Institute of EduCation on4 Research Conducted under Grant No. NIE-G-78-0005

.

. Patricia A. Taylor, Ph.D.Department of Sociology ,

University of Virginia

by

Burke D. Grandjean, Ph.D.'Department of SociologyUniversity of Virginia

. Principal Investigator: Patricia A. Taylor

Performing Orginization: Sweet Briar CollegeSweet Briar, Virginia

o

*Since grantees conducting research and development projects under Govern-ment sponsorship are encouraged to express their own judgment freely,.thisreport does not necessarily represent the official opinion or policy of theNational"Inttitute of Education. The grantee is solely responsible for thecontents of this,report." This report may be reproduced, in whole or in%part, for use by the Federal government.

'June, 1981

U.S. DEPARTMENT OF EDUCATIONNATIONAL INSTITUTE OF EDUCATION

EDUCATIONAL RESOURCES INFORMATION

CENTER (ERIC)/Tins document has been reproduced as

received from the person or organizationoriginating it

1 Minor changes have been made to improvereproduction quality

Points of view or opinions stated in this document do not necessarily represent official NIE

position or poky

a

SCHOOLING, TRAINING, AND PATTERNS OF OCCUPATIONAL CHANGEAMONG CAREER CIVIL SERVANTS, 1963 - 1977 .

.

ABSTRACT

This project examined the careers 'of a one percent sample of Federalcivil servants for the years 1963 through 1977. The areas of employmentinvestigated included economic returns to schooling and experience, factorsaffecting promotion and 'turnover, and occupational career ladders. A seriesof hypotheses and objectives specified at the outset were subjected tomultivariate statistical analysis using the official personnel-recordsiofover 68,000 Federal employees.

The results show that the main conclusipns of human capital researchin economics and status attainment research.in sociology hold for the Federalgovernment. However, internal labor market characteristics must also be con-sidered for a full unddrstanding of civil service careers.

, Theotexploration of career ladders concerns the extent to which-occupational change involves movement between-jobs with similar attributesand, by inference, similar skill requirements. Here the findings suggestthat mobility between major occupational categories is relatively infrequentin.the Federal services . and that when it does occur the attributes of originand destination jobs are not very similar. Attribute continuity is gredtestfor mobility involving a change of gradd or detailed occupation without achange of major categories, fo ;- employees in the'middle of their careers TSCompared to recent entrants, and for two of the -nine attributes considered --

the knowledge required by the job, and its physical demands, includingtechnical skills. The implication for educational practrice is that aconcern for teaching transferable skills to prepare students fora careerof occupational change should hot'sacrifice thordugh mastery of specificsubjea matters.

03

t

-3-

. TABLE OF CONTENTS

Title Page

Abstract

Page Number

1

2

Hypothesis 4 15

Table of Contents 1 3

List of Tables 5 .,

Introdyction 7.._

Organization of the'Report a T 7 .

Research Design and Data Set . 8

Human Capital Hypotheses 10

Hypothesis 1 12

,

H9:gothesis.2 12"

Hypothesis 3, 12

Hypothesis 5 T7

Hypothesis 6 2'2

AOMobility 31

HypothesA7

< Hypothesis 8

'Hypothesis 9

Hypotklesis 10

Hypothesis 11

31 `v

41

44 ...

-4,4

4

fl

. . . 44 2 ,A,

', V

o

Occupational Career Ladde'rs .,

46

Objective 1 . , 46*-,,-

.

Objective 2 46

. -

4

i

,-,

-4 J

Page Number

*

Objective 3 46

Objective 4 46

Summary and Conclusions\, ,

55'%

Appendix A (Hypotheses and Objectives)

Appendix 8 (Department of Labor Final Report)

Appendix C.(The Data Base)

Appendix D (Bibliography)

If,

Of.

I/

r, = ...S. 11.

.... e

,5

4

CO

I

-5-

LIST OF TABLES

Table 1: Education Level Attained, by GS Grade, and by

Rage Number

\Sex, August, 1974 11

Table 2: Pay,ttructure of,Minority/Sex Groups, for FourCohorts, 1963-1977, Predicting 1977 Salary 13

Table 3: Standardized Returns to Schooling and ExperiendeA,by Minority/Sex Group, for Four Cohorts,,1963-1977, Predicting 1977 Salary 16

Table 4: Effect of Degree 'Certification on Salary ofProfessionals and Administrators Only, for Four--Minorfty/Sex Groups, Net of Independent Effects,Cross-Section 1977 18

Table 5: .Regression' Equations Predicting 1977 Salary for theOn-Board, White-Collar Employees, by EducationalCertification . 20

. , . .

Table 6: Effect of Schooling on Duration and Expense ofJob Training,. Cross-Section, 1974 - 1977.,

Table 7: Effect of Training; Education, and Experience on

24

1977 Salary for Five Cohorts 25

Table 8: Employee Traiging by Minority/Sex Group, fdr AverageHours of 'Trai4ing, % Receiving Training, and Average-Cost of Training Per Worker Hour, Fiscal Year, 1976 . . . . 28

.Table 9i Regression Estimates of Minority /Sex Status inPredicting Differences in Cost of Institutional JobTraining with Non-Minority Males Among Federal,White-Collar Employees; 1974'-1977

o f

30

Figure 1: Schematic Model of Socioeconomic Achievement in'the Federal Civil Service . . . . 32

Table 16: Variable Means and Standard Deviations for FiveCohOrt1 of Career Federal Civil Servants

o

34

Table 11: Socioeconomic Achievement Among. Longterm Employees:

Net Metric Regression Coefficients for Pre-1963Entrats to the U.S Civil Service

$-'37 ,

Tabler12: -SociOecOknomic Aphievement,in the Great Society andVietnam Cohorts: Net Metric Regression Coefficients.for 1963-65. and 1966-68 Ent6nts to the U.S. MilService . . ,. . . . , ....... ... . 3B

Table 13: aocioeeonomic_Achievement in the,Nixonomics and.Watergate Copor;s: Net `Metric Regression Coef-'ficients for 1969-71 and 1972-73 Entrants tothe U.S..Civil Service 39

6

c

0

-6-

Page Number,

Table 14: .Effectt of Education,MinoriWSex Group and4 Occupatignal'Category on Termination for

Entrants and Mid-Career Civil Servants 45

Table 15: A Log-Linear Analysis of OccupationalMobility, 19.69-1977: Standardized Lambdas . . 48

Table 16: Selected 1973 and 1977 Occupational Categoriesof 4169 Clerical Employees (Percents) 50

Tableil,: Factors in the Factor Evaluation System 51

Table 18: Continuity Coefficients for Raw Scores andProfile Scores on Nine Job'Attributes,by Mobility: Entrants to the U.S. CivilService, 1963-1973 ' 53

Table 19: . Continuity Coefficients for Raw Scores andScores on'Nine Job'Attributes, by

Mobility: Mid-Careei U.S. Ctvil Servants,1965. 54

Table 20: Effects of Selected Job Attributes (Raw Scoresand Profile Scores)on'Mobility: Entrantsto the U.S. Civil Service, 1963-1973 %56

Table 21: Effects of Selected Jqb Attributes (RwScoresand Profile Scores) on Mobility:- Mid-CareerU.S. Civil Servants, 1965

4

-7-

A

FINAL REPORT:.,

.

Schooling, Training and Patterns of Occupational Changeamong Career Civil Servants, 1963 - 1977

NATIONAL INSTITUTE OF EDUCATION GRANT #78-0005

Introduction

A dominant themd in American education has been "If you want a betterIjob, get a better education." Indeed, there has been considerable evidencethat education does have a great impact on one's salary and the socioeconomicstatus of one's occupation. Research on-"human capital" (Schultz,1961;'-`Becker, 1964)and the monetary returns to schooling and training has foundthat schooling is.positively associated with annual salary and/or lifetimeearnings--suitably discounted.for earnings foregone and expenses incurredduring education (e.g., Hanoch, 1967; Hunt, 1963). Additionally, research on eoccupational achievement has, folind education to be of primary importance,especially for males (alau and Duncan, 1967; McClendon, 1975).

Occupational change, howevier, is a social and individual process lesswell, studied, although the broad outlines of occupational change hive beenspettfied in a handful of studies. The importance of occupational changecannot be-overstated. With roughly five million American workers changing job'seach year--approximately six percent of the paid labor force (Byrne, 1975)--occupational change has become an essential consideration in individualvocattonal planning, in parfonnel administration within organizations, and innational labor force policies. .Educators,have come increasingly to recognizethat adequate career preparation should develop a, capacity to work at severaldifferent occupations between the time of labor force entP'y ap that of retire-ment, and research is now progressing on transferability of vocational' skills(see Altman, 1976). Yet educational programs aimed at career 'preparation,whether through schooling or .through on-the-job training, require not only, aknowledge of the abilities and preferences of individuals, but also a thoroughunderstanding of the social and economic forces which constrain and facilitateoccupational change (cf. McKinlay, 1976). Desptte the considerable empiricaland theoretical attention which social scientists have devoted.to occupationalmobility, however, the process of job change over the course of a career remainspoorly understood. The extent to, whichrpareer options are indeed patterned or,structured, the forms of such regularities, and the influences which may e"

channel a series of job transfers along, into, of out of a given-pattern, have:not been adequately examined. Our.report aims to contribute to know-ledge on schooling, training, and occupational change.

4Organization ofathe Report

This report is organized into six main sections. First, we will 'discussthe research design and data set used in our analYsei. .,Throughout this section,

b.

I

a

-8-

we will indicate the sighificave of certain characteristics.of the Federal' '

labor market and those which -are measured-in-our data set.

The next three sections deal with specific hypothese and 'objectives

'detailed in-oui:sgrant proposal to the National Institute of Education (seeAppendix A). These hypotheses.a114 objectives fall into the areas of "Human.capital - formil and 'on- the -job training, ", "Mobility- promotions andturnover," and "Occupational career ladderS."

. . .. ..........

i.. A fifth section 4S a summary of our work for the National Institute of .

Education with comparisons of our findings to, the private sectbr,and to the ,

general literature on pccupatipnal change and transferable skills. The sixthmajor section is that of Appendix B, our final report to the Department oftabor on work which was erformed in co junction with our NIE project.

j..

, .

''':'. ,

.Research Design and Data Set

. t.

The research, we present in this final report is based on a two and one- ....' ''''1',

half year study of a one percent sample of the automated personnel records of

t Federal civil servants, 1963'-1977. Using Federal civil servants as a basisfor considering osccupational change has its problems asvel as its benefits.As an internal labor market,, the Federal civil service has characteristicssimila to other internal labor markets: identifiable-ports of entry, promo-tion ladders, protection from the exigencies of the external labor market: .'

These similarities suggest that a study of occupitionalchange in the Federalcivil service will proVide,information generalizable to other labor markets.

I However, the civil service is distinctly different from the, private sector inanumber of ways: its operations are regulated by law;'salaries are-set byCongressional action and are pegged to the private sector; some services itprovides are not found in the private.sector. This latter consideration meansthat certain job opportunities available to Federal employees are not available .

to priVate sector employees, and vice versa. t

,

Given these differences, as well ars others, it is important to establishthe similarity pf the Federal civil service labor market to other labormarkets.In this way, generalizations may be possible regarding the transferability ofskills betwein private and public employment, as well as the similarity, or

lack thereo , In occupational changes. To accomplish tht task, we first used

a human capital model to analyze the relationship of education, training, andearnings. -These results provide the basis fpr'a comparison of the Federal civil

, service with the private sector. Second, we investigate activity in the Fectral'civil service in promotions and turnover to gauge the extent of two types of'

.4 occupational change%- And third, we examine the career. ladders in the Federalcivil service lising'entry occupation, detailed occupational changes, and a-

\slbills analysis using the Factor 5valuatiOn System.

. For most of our analyses, we used data drawn from automatedrecords main-tained by the'Office of Personnel Mallagement. Since 1962, the Federal govern-

ment has taken' progressive steps to computertze its personnel records. It

,began.theqederal Personnel Statistics Prdgram (FPSP4, a longitudinal ,work.history file on a ten percent sample of Federal civilian employees. The aim

was to provide a statistiCal basis for work force analyses which would contrib-

ute terational personnel management. Over the years, changes in the automated

data system have improved the accuracy of the records while increasing both the

,-

..',. )......

,r9-

amount of 'information 9tored'oh vach%Nemployee and the number of employees in.

the file. In 1972 FPSP was superseded by the Central Personnel DataFile(CPDF). Like its predecessor, CPDF covers virtually the entire Executive

.

Branch, except for White Molise staff, intelligence personnel, employees of theTennessee'Valley Authority, and politically appointed heads of agencies.Included also are thetGeneral Accounting Office, the Government Printing Office,,the U.S. .Tax Court, and the administrative offices of thegfederal court system(See Schneider, 1974). Transaction histories are Row maintained.o4 100 percentof these employees.

Monthly agency reports, subrAtted to the Office ofiPersonnel Managementby each Federal,agency, have been merged With the CPDF to provide constantupdating of personnel actions occurring throughout the Federal civil service.In the report that follows, we draw from that compilation of data. The bulkof the original data analysis in_the chapters that follow it' based ontheofficial personhel records of a one percent sample of Federal employeei, dracnprimarily from FPSP and CPDF. Additional files, such as the Minority GroupDesignator File, the Retired Military File, and the Training File, were alsotapped for infqrmation to build our longitudinal file, referred to here as theFederal Career,File, or FCF. This composite data file:on over 69,000uals employed by the Federal government for any lerigth pf time. between January,

'1963 and June, 1977 was made available for our use by Pie Office of PersonnelManagement.' .

The information included on FCF is of considerable 'variety. For example,on each Federal employee a status record as of June, 1977 (or as of theirseparatibn from Federal employMent) includes information. on birth date, educa-tion, agehcy, geographic location, sex, minority grotto status, pay group, salary,veteran's status, length of Federal experience, and so forth. The dynathicanalysis in our'study is possible from theihformation contained on transactionrecord's for each employee. The number of personnel transaction'Tecords varies'by employee, and in principle has no upper bound. The minimum is one, forthose who had just entered Federal service in June of 1977 or shortly before,or whose first transaction in FPSP recorded their"departure from. the Federalservice. The observed maximum in our sample is 64 for a long-term.employeewith, quite obviouslyvfrequent_ transactions: The average is 8. Personnel ----actions- recorded include, but are not limited to, grade increase, occupation ,

change, merit pay increase., salary adjustment, demotion, geographic change,

agencytransfer, furlough,,military leave, and separation, with codesAiavailableto distinguish the reasons for a transfer, separation, or furlough. -Personnelactions can also include A change in status, such as from career-donditionalappointment to career, or from excepted service-to competitive. Many of theseterms will be discussed in later sections. Readers Who mist) more detailedinformation should consult the Federal Personnel Manual, Processing PersonnelActions, and Personnel-Data Standards, all published b the Office of PersonnelManagement (formerly the U.S. Civil Service Commission . ,

o

The final portion of (CF consists of information on job training by'employee from 1974 through,1977, and includes cost and hours of training,purpose of training; type of training, and other related data. We are able,by the manner in which these various data elements were merged into FCF, tolink dethographic characteristics of an individual with a history of personneltransactions and training instances..

1 Q .

9

-10-

Forourresearch under)this grant, we used the data from the FCF to assess'the effect of various individual worker characteristics (such as education,years of experienpe, etc.) on lary using least squares regression analysis.Specific characteristics, or va iables,,of interest were, operationalized andtheir effects. on salary or turnover were assessed net of the effects of other.confounding characteristics suchbas age, sex, and race. While there arecertain limitation? in this method of analysis (the specifics of which are':discussed later), this method of analysis follows in the tradition of humancapital and status attainment research; as well as being a suitable methpd.tomaximize the detailed information in our data set.

.

An additional source of data used for the objectives in'this report wasgenerously.Supplied.by the Personnels Research and Development Center (1974)...

.

The Factor Eva)uation Systemois a technique designed to assist 'inproper blassi-ftcation.and grading of Federal civilian General tcheslule-jobs, which are non-supervisoryin grades 1 through 15 (where,18 is the top grade). ,Through.several steps of selection, evaluation, and 'wejghting, a, set of first five, andthen nine, job faetbrs was -defined and tested for use in position tlassifica-tion. Some-af.these-factors represent, we believe, skills in deal-ing,withtechnical information, other Teople, and communication. We will use thesefactofs as independent indicators of'skills to assess the relationship ofvarious skills.to types of occupational change. The reader should be cautioned,'however, that our skill measure inheres.ln the occupation of an individual, andnot in the individual per se. That is, persons are assumed to have high krtow-ledge mastery if they are located in an occupation which requires such mastery.Hence our measureof skill levels are only approximate._ We shall return tothis problem in our discussion bf tcupatidbal career ladders.

.

- HUMAN CAPITAL - FORMA,AND ON-THE-JOB TRAINING

In.the human capital tradition, education holds a prominent position inthe determination of,wages. Arguing that the.main determinants of productivityare cognitive abilities and technical skills, proponents of the human capitalperspective predict that both societal and individual income differences willreflect differences. in investments'to develophese abilities and skills.Consistent with the prediction, a number of studies, focusing on individuallevel data have.shown years of schooling to be positively associated with annualsalary end/or lifetime earnings--suitably discounted-for earnings foregone andexpenses incurred during education (e.g., Hanoch, 1.976; Hunt, 1963. Thoughsomewhat reduced, the association holds when measures of individual ability arecontrolled (see articles and reviews in Solmon and Taubman? 1973).

The Federal government is,nb exception to these general findings. ,As thedata from table 1 show, the higher General_ Schedule grades (4.e., the higherpaying positions) tend to have persOns with a bachelor's degree or higher as .

occupants. For example, of all GS 14-15's, 78 percent of the 'ale employeeshave at least'a B.A. degree, while the same is .true of 77.6 percent of thewomen in the GS 14-15 category. In.the GS 16-18 grade group, we find a similarpercentage of male and feMale'employees having high levels of education. As

might be e*peCted, the percentage of employees who have graduate schboling ishigher in these grades than in any'other grade grouping.

I '

1 14

Table 1: Education Level Attained, by GS,

r Grade,and by Sek; August, 1974a

GS GRADE GROUP

Education Leve140 GS 1-4 GS-5 -8 GS.9-11 GS 12-13 GS 14-15 GS 16-18

Males Females Males' Females Males

.

Females Males Females Males

,

Females4 Males Females

Less than H.S. 18.4% 7.2% 10.4% 4.0% 3.7% 1.8% 1:1% 1.0% 0.4% 0.3% 0:2% 2.0%7

U.S. Graddate. 37.8 47.6 .31.2 37.8 17.7 22,.3 1_0:6_ 12.8.

5.6 6.0' 2.9/

.1.0

-Training Beyond. ,

U.S. ,. 11.7 18.7 11.8 1.2 11.7 13.4 6.3 9.2 ' 3.2 4.7 1.5 5.0

'

-a1 Year College 17.7- 16.9 16.0 19.6 14.8 16.6 9.8 12.3 5.2 5.6 1.9 2.(:

2-4 Years College 10.1 7.2 12.2 9.7 12.4 16.2 -10.8 9.3 7.6 5.7 4.8 4.0

Bachelor's Degree 3.i 1.8 13.6 5.8 20.9 16.9 23.5 ,18.3 19.5 10.4 14.6 10.9

Graduate Study 1.1.

0.6 4.8 2:0 18.8 18. 37.9 37.1 58.5 67.2 74.1 75.2

100.0 100.0 100.0 100.0 100.0- 100.0 100.0 100.0 100.0 100.0 100.0 00.0

12

aFrom a 100T couk.;---:-Source: Q.S. Civil Service Commission

-/

.Conversely, very small percentages of employees in the highest gradegroupings have only a high school diploma or less education. Indeed, for thecategories of less than high school and high school graduates, we find that thepercentage ,distfibution of employees across the six grade groupings consistent-ly decreases from the GS 1-4 to the GS 36-18 group, with only one slight

ception. These data alone suggest the importance of education in the grade

'stributionof:employees. On the other hand, that some employees are in the

'ianagerial-and adoWnistrator grade groups withOut a 'collegadegrae suggeststhat it is possible to perform administrative functions without formalized`higherJeOucation. That is, over twenty percent.of all GS 14-151-s have attainetheWpokSt'ons without having what is generally regarded as a necessity forentry into the managerial positions, College degree. Thus, forces other thaneducational credentials affect career success in terms of,General Schedule

.

grade. The work experience of an employee is no doubt important in increasinghis or her value to,an.employersince many work skills can only be learnedthrough experienCe: In the case where a work process is learned only by doing,rather than by schgoling, an employer might very well rate experience higher

) -than education and promote employees. on the basis of seniority. Thus, 4n-the-.job training or experience call be substituted or actually preferred to educa-tion for, certain positions.

Our first set 'of hypotheses deals directly with trying to estimate theeffects of education and experience separately. We hypothesized that:

H(1) Yea?i of school and amount of experience will bepositively associated with salary.

H(2) The association jn H(1) will be. stronger for whitemales than fin---f.emaleS and minorities.

H(3) The highest returns to schooling will be for those.employees whb have high leyels of education and high levelsof experience. .

The results from our analyses which attempt to examine- these hypothesesare presented in Table 2.

, .,

eachused regression analysis to predict salary for each minority/sex group

in each cohort., Presented in Table 2 is an estimate of the pay structure ofeach group for each entry cohort. In short, ths,b-coeffftients for each of theindependent variables used to predict salary is an estimate of the averageeffect of that variable in determining the salary of a given employee (seeKluegel, 1978;.Tdylor, 1979) . The pay structure thus determined can be com-pared across cohorts and across minority/sex groups for changes over time byparticular indepepdent effects.

Given the emphasis of the government in using'merit criteria for employ-ment, we would predict the following. First, we would expect given pastpatterns of hiring and advancement that education, while positively related tosalary for all groups .[H(1)], has a stronger effect on salary for white malesthan for any other group [H(2)]. This is due in large measure to the use ofeducation as a selection instrument for white males, and the exclusion of womenand minorities- from higher-paying occupations regardless of their education.Second, however, the effect of education on salary for white males should be

14

4

TABLE 2 : PAY STRUCTURE OF MINORITY/SEX GROUPS, FOR FOUR COHORTS,1963-1977, PREDICTING 1977 SALARya

Independent:

Vartables

WhiteMales

EducationExperienceExperience2Entry AgeEntry Age2D.C.

Disabled Vet.Other Vet.R2 (adjusted)N

X 1977 Salary

EducatiOnExperience

White Experience2Females Eptry Age

Entry Age2D.C.

Disabled Vet.,Other Vet.R2 (adjusted)N

X 1977 Salary

Cohorts

1963-65 1966'48 1969-71 t 1972,73

1398*** 445*** 897 * ** 984***35 159*** -209* ' 163***

.12 -.67* 1.93** -3.220'179 850*** 219 932***

-1.26 , -11.41** .31 -12.00*4417*** 4327*** 4484*** 3387***-62? -940 -26344 -304-820 o. -27 895 -721

.416 :457 .473 .444300 387 193 254

$23,306 $20,788 $18,844 $15,628

1503***-43

.29

188-3.88

3947***10 .916

.51A

169$13,700

1572***25

-.02-45.81

3018***-2971

47

,467

252$12,641

1387***79

-.18

248-2.623525***

-530.573

1004***68 * **

-.65318*

-4.62*

-769.406

105 Ar 237 -

$12,611 $10,688

Education 203a*** 1929*** 929 221**1Experience 274 4670 17 121*

MinorityMales

Experience2Entry Age

-.86

651

4.310-2119*

.73

1502

-4.29'

9780

Entry Age2 -9.70 414.92** -22.29 -12.89

D.C. -1859 . 369 1 4853 467

Disabled Vet. 422 3062 -1038 -2092

Other Vet. 845 -1044 1595 -2895*

R2 (adjusted) .470 .531 .190 .427

N 22 43 23 63

I 1977 Salary $18,897 $16,889 $15,l&5 $13,908

Education 1576 '991*** 1280*** 664***

Experience -106 94 46 -23 .Minority Experience2 ..57 -.44 -.11 .71

Females Entry Age -235 .472 561 69

Entry Age2D.C.

2.31

18030

9.610

1472-8.97-905

-.97

368

Disabled Vet. -67 -4586

Other Vet. 936 -1145 -484 -1534#

R2 (adjusted) .362 :269 .331 .640

N 48 100 59 78

X 1977 Salary $12,778 $11,602 $11,592 $9,723

a a: Levels of, statistical significance on this table andon all subsequent tables are; s4opf.-.1; *, ps.05; **, ps.01;

-and ***; p$.001.

15

,

.>

greater in the-older cohorts [H(3)]. However, as more and more minorities-andwomen are eilected for employment and placed into ocoupationssimilar to whitemales, we would expect that education increases in its effect on employee'ssalary. Therefore, weAmight observe a different pattern'for minorities andwomen as compared to white males.' Finally, the explained variance in salaryshould be higher for whit p males than for any-other group since the use ofmerit .criteria has been applied to differentiate mainly among that group ofemployees to the exclusion 'of others. That is, human capital variables suchas education and experience shoUld do more to explain variations in salarywhere they have been instrumental in determining access to jobs and promotions.

The returns to the'human capital-variables themselves display a surprisingpattern. Both education and experience have generally positive effects onsalary [4(1)]. The effect of education on salary becomes greaterNthe older thecohort [H(3)], but the effect of experience declines over the-four cohorts.These patterng are generally observed for four minority/sex groups.However, the effect of education on 1977 salary is generally lower for whitemales than for any other minority/sex group [H(2)]. Thi4 suggests that educa-tion may be-used differently for different groups of employees. For example,education can be used as a screening mechanism for entry, where minorities andwomen are less likely to be placed into the good jobs. However, educationmight also be used internally as a selection mechanism more for women andminorities than for, white males. That this may be so can be inferred froth anunderstanding of career lines. If white males tend to have more predictablecareer lines in which they move on an established career ladder, educationwithin the labor market may not be as important in differentiating employees.If, on the other hand, women and minorities can use their educatiqn to jumpoccupational divisions for increased4talary, then education would have a greateraffect on their 1977 salary than on their entry salary relative to white males-

However, experience does not seem to have a pronounced effect on 1977 -salary within the cohorts, and this pattern seems to hold for all four minority/sex groups. Basically, what is demonstrated here is that the effect of exper-ience on salary within a cohprt loses its importance, within the first fewyears. In essence, the difference in accumulated skills between an employeewho has worked 14 years versus one-who has worked 10 years is considerably -

smaller than the difference between an employee who has worked 5 years comparedwith an employee pf only I year. Similarly, entry age has in general apositive affect on 1977 salary suggesting that experience outside the Federalgovernment contributes to earnings. However, the,effect is strongest for themore recent cohorts.

Locationin D.C. is especially important in the pay structure of whitemales across all four cohorts. Being in the home office of the Federal'government increases their salary by $3,387 for the newest cohort to approxi-mately $4,400 for the earlier three cohorts. A similar effect, althoughconsiderably smaller, is found for white women. This pattern 'is consistent withwhat we know of the structure of careers. Location in D.C. has a larger pay-offfor the older cohorts since: (1) the highest paying jobs in Federal employment'are located in D.C., and (2) most recruits are not hired directly into thesejobs but must move through the civil service for some period of time.

However, neither group ofminarity employees has a consistently positiveeffect froin employment in D.C., Given,the recruitment pool for the D.C. area.

iU

-15-

this finding is understandable., Whites who are;in D.C. may.have,moved therefor a promotion; hence, locftion in D.C. is only associated with higher,salaries rather than causing higher salaries. On the other hand, the laborforce of the D.C. area is approximately 70 percent black, so that minoritieswho are in Federal employment in D.C. were more likely to have been recruitedinto lower level jobs in the D.C. area than were whites.

Finally, the effect of veterans' preference variables on salary is mostlynegative; 19 of the)28 coefficients are negatively signed, but only threeapproach statistical significance in the two-tailed test. The fact that,the

: preference is generally negative may suggest that persons of lower ability arebrought into the Federal government and do not.compete as well with other

''employees._. This seems true of all groups of employees, notonly white males.Given the lack of statistical significance, however, the effect of veterans'preference on salary is'not very pronounced. Therefore, it appears that thepreference is more likely to have effect in retention rights during reduction-in force, and preference for a claim on a job, rather= than in the promotibnand salary of employees once on-board the Federal work force.

In summary, what stands out most clearly from an examination' of. the paystructure of the four minority /sex groups for 1977 salary is that our expecta-tions of.returns to education and experience work best for white males and areonly somewhat applicable to women and minorities. This is.not to argue thateducation and experience do not affect the salary of minorities and women.Rather, the importance of such variables is qualitatively affected by the sexand race'o the individual employee.. Such effects are probably mediated by.occupational stream, and are also variable across the timing of the career in.Ways different froril white.males..

However, we have dealt with only the unstandardized effects of educationand experience on salary. It could be that since both education and experiencehave different mean values, and different standard deviations, that we shouldexamine the standardized effects of these,two variables on salary. Hence, wehypothesized that;

H(4) Returns to schooling and to experience will be similar inmagnitude. ° %

Presented in Table 3 are the standardized estimates (the beta's) of the effectsof education. and experience on 1977 salary,'controlling for the same independ-ent-variables as in Table 2. The standardized estimates adjust for differencesin the measurement of the two variables, so that we might better compare themagnitudes of.;the effects of educition and experience on salary.

As is evident frbm.the table, very w coefficients for education andexperience are similar :In magnitude. Edu tion generally has a greater effecton salary than does experience, for all cohorts and inority/sex groups withonly two exceptions. Indeed, as we noted firm Table 2, the coefficient foreducation.nearly always attained statistical signifi ance whereas the coeffic-ients for experigggararely did so. The explanation 'for.this'finding may be _

that we have eliented a good deal of the explanator power of experience bybreaking our.sample into cohorts. Most studie,s in th human capital traditionlook to the cross-sectional effects of experience, and find stat4sticalsignificance. Indeed, in some of our earliest work on his project we found

-17

-16p.

Table 3: Standardized Returns to Schooling and Experience,.by Minority/Sex Group, for Four Cohorts, 1963-

1977, Predicting 1977 Salary.

Minorgy/ IndependentCohorts

Sex Group Vartablesa 1963-65 1966-68 1969-71 1972-73itc

White Education .525 .580 .346' .461Males Experience .126 ..576 -.584 .257

White Education .549 .630 r° .608 .580Females Experience -.224 .118 .252 .217

Minority Education .748 .656 .368 -.501Males Experience .740 1.452 .072

..231'

Minority-,

Education , .551 .437 .576 .754Females Experience -.678 .639 .172 .163

a: Net of additional effects ovariables included in Table 2. R2is and N'sremain the same as in table 2.

k

PI;,+ ".".ei

1,

,

experiehce to be significantly related to salary 'n ,the cross-section byminority/sex. group (Taylor, 1979).

Nevertheless, the_ results from Table 3 suggest t .t education has gene.47 ,allia stronger effect on salary than does experience. 'art of this effecteCould be due to the occupational group that an employee i hired into; that[is; education may have an effect not only because of thelmportant skillslearned in the educational setting, but also because of the degree required bySome occupations for entry. We therefore hypothesized that,

H(5)' Controlling for years ,of schoolingand government experience,earned degrees will be positively associated with salary.

.

This hypothesis was somewhat modified in the:course of our study. Rather .

than focus on the number of,earned degrees (such as B.A., McA., etc.), we focusinstead on whether the occupation an employee has entered requires a particUlareducational'credential, such as a B.A. in psychology, an M.D., etc.

The Federal government4provides two sets of standards for job and employeeanalysis. One standard for job evaluation is the Classification Standards, inwhich the duties, responsibilities, working relationships; technicaletc., for persons employed in a particular occupation at a given level (i.e.,job) are described. This is somewhat analogous to'the demand side of emarketanalysis. The "mirror" to the Classification Standards, or the supply side Ofa market analysis, is the Qualification Standards, which describe the experienceand/or educatron an individual must have t qualify for entry into a job. .Thesestandards report whether an individual Imus have a degree in ,a partictilar field, .

or whether experience can be substituted for some or all of the education. Wewill use this information to analyze the-credentialing hypothesis put forth byBerg (1970), Collins (1979), and others.

Essentially, the view expressed by these writers:is that different levelsof.education,do not necessarily differentiate levels of skills. Rather,different levels of education suggest the possibility of trainability to an

employer, the preence of certain values such as deferring gratficatibn and )

future time orientation, or even native ability. Hence, an educatton credentialis worth more to an employee than the sum of the years of schooling taken to getthe degree:

When qualification$ for job entry are as highly specified as in the-Federal civil service, the researcher is able'to determine the degree of edu-cational certification required for hiring and placement. Uting the Quaifica- ,

tions Standards'as a guide to the presence of'a requirement for eludattknalicertification, four groups of white-collar occupations were design00*aroughly:lepresenti*different degree requirements. The first two Ponrefet00*.CertqUation 1 and Certification 2, reflect the fact thatpersp,:,t tenter Job" a given',occupattonal title must have an educationaldegreC1), itlield (see footnotes a and b, Table 4). :E0se jobs are'primar-ily in'ther dical; 'legal, and engineering fields. However, since some medicaljobs are 'highly sex-typed (e.g., nurses), as well as therq being both a sub-, .°

stantial difference in length of'training and large salary differente be#ennurses and physicians, the Certification 1 group represents the more mad`;__

'dominated Oftupations which require certification for entrance, while:Certification 2 the generally recognized female-typed occupations.

4

f.

fr718-

o #

Table 4: Effect of Degree Certification on Salary of Professibnals,and AdministratOrs Only, for Four Minority/Sex Groups, Net

,

of.Independent Effects, Cross-Section 1977.

4.

-41

Min ority/Sex Groupd

CertificationGroup

,Non-MinorityN Males

.Non- Minority

FemalesMinorityMales

MinorityFemales

N (4',246) (1066) (394) . (251)'

Certification )a 2,820*** 4,984*** 4,300*** 4,496*

Certification 2b -3,721** - -787* -4,106# ' 4,473

EducationSpecifiedc -487* 818* -0 -303.A

9

a: Occupations in the Certification.1 Category in the WhiteCellar Work Forceinclude: Safety Officer, Chaplain, Medical Officer Pharmacist, Optometrist,Podiatrist, Dentist, Electrical Engineer, ElectronictIgineer,Aerospace Engineer,Marine Engineer, Laval Architect, Mining Engineer, PetroleuF Engineer,Agricultural Engineer, Ceramic Engineer: Chemical Engineer, Industrial EngPieer,General Attorney. Veterinarian, Engineer, Safety Engineer, Materials Engineer,

Architect, Civil'Engineir, Sanitary Engineer, Mechanical,Eliginer, NUtlAearEngineer, Biomedical Engineer, Tax Examiner, Administration Law Judge,Deportition and-exclusion, Adjudicating, Patent Attorney, Tar Law Specialist,Metallurgist, Educational and Vocational Trainer, Aircraft Operation, ,FirePreventioil Engineer, Welding Engineer.

b: Occupations in Certification, -2 include: Nurse Anesthetist, Nurse, PublicHealth Nurse, Occupational Therapist, Physical Therapist, Corrective Therapist,Manual Arts Therapist, Educational Therapist, Medical Technologist, SpeechPathology and Audiology, Dental Hygiene, and Education and Vocational Training.

E: Occupations which Require Considerable Training in h Given Field, but:notNecessarily Certification or,a Specific Level of Education include: Economist

4

Psychologist,,Sociologist, Social Worker, Archeologist, Co-:outer Analyst,Communication Specialist, Microbiologist, pharmacologist, Zoologist, PlantPhysiologist, Entomologist, Botanist, Range Conservationist, Soil S'ci'entist,Agronomist,Firial Maitager, Accountant, Budget Ad-linistrater, Dietician,Medical Researc Librarian, Landscape Architect, Financial Analyst, Realt$: Series,Patent Advisor, Health ,Physicist, Geophysicist, Hydrolorist, Physicist, Cilemist,Astronomer, Meterologist, Geologist, Oceanographer, Cartographer, Geodesist, -

Librarian, Actuary.

'd: Net of the effects of years of schooling, age,and age.squared, veteran's preference, Federal experience andexperience squared, position occupied,, supervisory

.

status, and location,inD.C.

114

,

4,- tIr- n4tie ..,-, _ ,,third occupatio al group cons sts of tholKocUlax,i4h ati.which the educe-

it onal field is spec ied-(Zdfic tion Specif.is'not required per All of er profess`,

. are included in the fourt ategory:

ut egree in that fieldtive occupations

In Table 4, we present thee ffects of4ctminority/sex groups, controlling fo 'nine inde enanalysis used three dummy.variables ( tifidatio40Education Specified) as predictors on 197 salari.14certification variables is that group of professio 1

pations which,require no particular education to f:these positions there is a general preference forpe

diary for four'ables. The regregsionTfication 2, and

d -'t category of thed adm atNe occu-

al ough 4 hi-rIng for

'-011thle-bach8lors.'.

0 0

degree. .The coefficient for each certification grodgon, is the difference-fn average salary betWeeh that group and the deleted caTigoty.

.

For minority males, for example, those persons who are in occupations whichrequires degree in the medical fields, engineering, law, etc.-, make $2,820more in salary than do those persons in the "Other Professionals and Adminis-trators," the deleted category. Similarly,/white feniaTeS'imake almost $5,000more in the Certification 1 group than do females in other occupational groups.The same is true for minority males; however? minority females do almost aswell ift the Certification 2 group. The Certification 2 group representsoccupations whidh require a particular degree, but which are generally'female-

.

typed occupatiohs (nurse, medical technologist, etc.). Hehce, minority femaleswho might:have very litited opportunities in the "Other Professionals andAdministrators" group, could do better in aq occupation where entry isthighlyspecified,, even if it is poorly, paid relative to male-typed-occupations. =,

,.-

___.2....) ..0 .

Another way to examine the effect of degree certification on salary is tocompare the effects of predictor variables on salary within certifidationgroups. The-results from regression analyses predicting 1977 salary withineach educational group are presented in Table 5. , The unstandardized regressioncoefficients for each of thirteen indepeldent variables are given, along withtheir significance levels, the explained variance, and the elasticities-9otboth -Years of schooling and experience: The earnings faaction far each,_educa-tional group explains alarge amount of the variance in salary, from .517 to.613, thus indicating again, a good Sit of these variables.

In he educational group of Certification 1, the analysis includes thoseemployees in occupations which involve-a lengthy course of study and in whicha degree is required for entry. Even though a considerable amount of the pqier..of education should be used in explaining entry into these occupations, thereis nonetheless a considerable effect of years of education on%salary. Thevalue .of schooling within this educational group is approximately $1134 peryear of schooling. The elagticity coefficient which-indicate& the percentagechange in the dependent variable for a-percentage change, in the independent'variable also suggests a strong positive effect of schooling on salary. Indeed,a one percent change in schooling is associated with a .736 change in income4.

.. , e. 'In the Certificatioh 2 group_,_ years of schooling also shows-a strong

effect on 1977 salary. However, the schooling coefficients from Certification1 and Certification 2 do not differ by a statistically significant amount.-This may in part be due to the truncated salary range forphysicians intheFederal government. Medical doctors enter at d rather high salar4leVeli but

.A,1 .

21

Vs.

j Table

-20-'

5t Regression Equations Predicting_1977 Salary for the On-Board,

White Collar Employees, by EducationalCertification

Years of Schomling

(=Elasticity)

Experience

(Elasticity)

Experience2

Age

ge.2

Disabled Vet -

Other Vet

Supervisory

Position Occupied

D.C.

Hinority Male

Non-Hinority Female

Minority Female

R2

(adjusted)

N ,

saiiry

i Education

Certification la Certification 2b

Educationc Other Professional

ii34***'

.736*

311**

.161 41_

1/87r1,..

-10.83***

,

.1t--2488*

T1818***

-311

3500***

1736***

-510

-2144*

-2897

973,

$26,729

17.4

esk.f

rt*

Specified and Administrator

1096***

1.05

306***

4.

.187

-2.82

112

-1749

-327

=129

2436***

3207***

1,51?

-957*

.643,

440

$16,1104

15.4,

825***

.608*

650**

996***

.705

'480* * *

.429 .397-

-10.05" - 6.77 * **gt** ***

718 762*** ***

-6.91 .-. -7.55**It'

**-18741r -1841

-1148**t, =.617**

579***,.

.%% .415***

'310***

-1222*

-2727***

-4313***

.543

1025

$22,167

-16.3

4958***

-1762***

-4117***'

-4728***.

.454

3541

$20,901

14.8

a: Occupations in the artification 1-"Citegory in the White - Collar Work grceoincludp: Safety Officer, Chaplain, Hedical Officer, Pharmacist, Optometr4st,Podiatrist, DeAtist, Electrical Engineer, Electronic Engineer, Aerospace Engineer,Marine Engineer, Naval Architect, Mining Engineer, Petroleum Engineer,Agricultural Engineer, ceramic Engineer, Chemical Engineer, IndlAtrial Engineer,General Attorney, Veterinarian, Engineer,. Safety Engineer, Materials Engineer,Architect, Civil Engineer, Sanitary

Engineer,...Mechanical Engineer, NuclearEngineer, Biomedical Engineer,' Tax Examiner, Administration Law Judge,Deportation and'Exclusion, Adjudicating, Patent Attorney, Tax Law Specialist,Metallurgist, Educational and Vocational Trainer, Aircraft Operation, FirePrevention Engineer; Welding Engineer.

b: Occupations in Certification 2 include: Nurse Anesthetist, Nurse, 4,b lieHealth /Furs", Occupational Therapist, Physical Therapist, Corrective Therapist,Manual Arts Therapist, Educational Therapist, HedicalTechnologist, SpeechPathology and Audiology, Dental Hygiene, and Education and Vocational Training.

c: Occupations which Require Considerable Training in a GiNen Field, but notNecessarily Certification or a Specific Level of Education incllde: EconomistPsychologist, Sociologist, Social Worker, Archeologist, Compuelm Arilj?styCommunication'-Specialist,Hicrbbiologist, Pharmacologist, tdologist, PlanPhysiologist, Entomologist, Botanist, Range. Conservationist, Soil Scientist,Agronomist, Financial Manhger, Accountant. Budget.Administutor, Dietician,Medical Research Librarian, Landscape Architect, Financial Analyst, Realty Sefies,Patent Ad;isor, Health Physicist; Geophysicist,

Hydrologist, Physicist, Chemist,Astronomer, Heterologist, Geologist, deanographer, Cartographer, Geodesist,Librarian, Actuary.

e 4

CI

-2i-

.because of the,structure of the Federal pay system, their salary is not nearlyas great as in the private sector. Hence, many highly specified M.D.s arelikely to leave Federal employment for more tucrative positions elsewhere eventhough attempts have been made to increase physicians' salary above the white-collar pay schedule.

For. those, occupations in which education \,s._ highly specified but a degreeper se isolot'required,'the return to each year of education is approximately925, with an.elaiticity of .608. And, in the educational group where thereis much less educational specificity, the return 'to an additional yea'r.of'schooling is $996, with an elasticity of-7705. In each educational group,years of schooling has a considerabletffect on salary- even though a credential-ing effect has already been substantively controlled. Moreover, the b coeffi-cient for years-of-schooling in the education specified model diffiers.signifi-canfly from every other schooling coefficient.

While these results are in no way conclusive, they do suggest that there. is a conS4derableeducation effect' over and above a credenti'a'l effect across

sever 1 educational categories. We have net examined here the relative effectsof ab lity, metivation, nor type of schooling, all factors which could affectthe r turns to schooling. Ndnetheless, it is quite possible that measures ofability would diminish the credentiil effect as well as the effect of years ofschooling in any subsequent analysis. Thus the relative effect of credentialv,s. years of schooling on salary would remain fal-FriEable.

The additional human capital variables yield relationships with salarywhich have been discussed to some extent earlier. In summary, both age andexperience have positive earnings slopes, though the.rate of returnto theiecharacteristics declines over time. Disabled veteran's status has a greaternegative coefficient associated with salary than does the other veteran category,.

AP".For the Certification 1.group, being in the excepted service has a strong

positive.effdct on salay. Thisis due.notOubt to the'fact that many lawyers,for exampl, are hired into excepted:positions due to the political nature oftheir work. Or, as is the case with physicians, salaries are elevated in manyexcepted positions so that the Federal government can better compete'with th'eprivate sector in retaining physicians. Similarly; in the Certification 2group, excepted service is-positibely related'to salary, For the'other two

'educational groupAr hOweveri positions in the excepted service have a negative.'relationship with salary. And as with earlier anal;es, location in theDistrict of Columbia is quite beneficial for all the educational groups.However, the difference between D.C. 6d field status for the Certification')group is somewhat smaller than for the other three educational groups.

.4,

Of special note in the analysis'of the credential versus education argumentis the relative effects of thinority/sex status on s Lary. For example,Doeringer and Piore (1971:141) have argued that, "i general, internal alloca-tive rules, being less nebulous than entry rules, annot be as readily adjustedto mask continuing discrimination." Stated in another way, given the specificityof rules within an internal labor market regarding hiring, promotion, transfer,etc., there should be less salary discrimination than in non-protected labor .markets. Taking this reasoning further, a plausible argument could be madethat within a BLM, the-greater the specification of occupational requirements,

40 23

k

- 2

-' the less the salary discrimination. Indeed, Blalock's theoretical propositionson occupational discriminationContain related argumenq (1961:245-246). Forexample, the greater the competition among employers for persons with highperformance leyels, the lower the degree of. minocity.discrimination by em-'ployers." And, to the extent that it is dif4cult to prevent the minorityfrom acquiring the necessary skills for high performance, the lower the degree, kof discrimination." It it generally argued by employers that "qualified"minorities and women are difficult to find- Therefore, if the arguments ofDoeringer and Piore, and Blalock are. valid, minorities would be hired and well-paid in those. occupations where *there is greater cOMpetition among employers.Given the results of the regression analyses, it.appears that this line ofargument seems to hold for. minority males. In those occupations which have themost highly specified educational entry reciuiremedts, and where there,is con-siderable competition from the pri('ate sect*, minority males have salariesnot significantly afferent from.white males. Even in'those educationalcategories where there is less eduCational specificity,' minority males do notlag far behind white males. More'importantly for this analysis, the lessspecified the entry educational requirement, the greater the salary disparity..

For Women, however, the results still suggest some salary disparity forall four educational groups. Both minority and non-minority women lag farbehind white males in salary, 'from $2727 to $4728, with the largest salarydisparities found in the group with the least amount of educattiOnal speEificity.While theattern of findings is not completely linear, it appears that both

-- groups of women are better off vis,a-vis white males in occupations wherethere is a high degree of educational specifitity: These findings` againsuggest that the administrative rules can be used to reduCe labor marketinginequality.

Experience ,and Job Training

There are additional types of learning which may be as important to anorganization and thus to the career of an individual within the organization.Generally, it is believed that experience On-the-job is an informal way oflearning, and this assumption forms the basis of much.of the research inon-the-job training (see Mincer, 1962, discussed below). Experience at aparticular task should facilitate productivity as a worker becomes familiarwith a task, and therefore the worker needs less supervisory guidance, makesfewer .mistakes, etc.

Also important, however, is formal job training. This training is formal.in the sense that it is giveW in a classroom environment, with a specifiedinstructor, teaching materials, and hours of training, and it is utually apartfrom tbe Work site. We refer to this'type of training as institutionalized 4

job tfaining (or IJT) as distinct from on-the:job. training (OJT).4i

Very little is known about IJT, asMost research to date has concentratedon returns to experience or OJT. However, the Federal Civil Service Commissionhas kept records on formal job traini g received by employees beginning in

. We hypotheized that,

41

H(6) Duration and expense of j training will be positivelyassociated with years of school n .

2,1

-23- -

Using the training icords provided by CSC, we estimated the. effect of educa-tion on 'hours of I3T received and cost of$!JT. We 'reasoned oat the greaterthe amount of schooling, the more likely that there would, be training programsto further advance and fine tune an employee's skills. Persbns with highlevels of education would show "trainability" to the employer, and would',probably be in field's which yequixed- even more Specific trallOW

dItl-

Pregented in Table 6 are;ttleAgults of our analyses. We included in ourstudy those persons who were On:61:00in 1974 regardless of when they entered,and were still onboard in 1977. *Given some exigencies with the data,.we usedonly training records for 1975 and 1976. Therefore, the effects presented inTable 6 represent the effect of schooling on IJT received in 1975 and 1976.Three models are'presented. In Model',1, we estimated the effect of schoolingon training without additional controls, and as we hypothesized, there was a,substanttal effect. For example,,for every. year of schooling an employee hark,s/he received, on the average, $2S.91 in IJT in 1975 and 1976.

In Model 2, we estimated the effect of schooling on receiving trainingcontrolling for age and-experience. Again, we found substantial educationeffects, although these were somewhat reduced from Model 1. That is, the moregdOcation a person has, the more likely s/he is to receive training.

However, when we controlled for'occupatianal .Stream (Professional,

Administrative, Technical, Clerical, and Other) by using a,set of dummy'varia-bles inthe regression analysis, we found the effect of education on trainingcompletely eliminated. This suggests to us that education is important indetermining the amount ortraining'Ofte receives, but its effect is lmOst c6m- -

4. pletely indirect, operating through occupational stream. That is, educationmay determine one's - occupational, group, but it is,occupational\group which in'turn determines the amount of .IJT an employee'receives.'

t.An additional question whieLis important to us,-is the eltdnt to which

training affects salary.of an employee. Essentially, do the skills,learned itraining, or the ct that someone his been to a training course have an effecton his_.or hen alary -

In Table 7, the results from regression analyses predicting salary fromLIT and other employment variables which would likely affect the relationship oftraining and sal'Ary are presented for five^cohorts of employees. The amount of

4, job training received by an employee is measured here by the colt of trainingduring1975 and 1976,.the only years for which reliable training data were,available for this analysis. The training data were used-over the two yearssince some career development plans might call for a serios of courses spreadover considerable time. SalarY,is then regressed onto cost of training, years:of schooling, age, length of goTernment experience, minority/sex group, occu-pational group, and pre-training salary separately for each of five cohorts.The explained variance is reported for each-'of the regression analyses, and itvaries-between,R2.. .840, to R2 = .920. The high R2.is.due'in large measure to -

--the control variable o pre-training salary. In effect, salary in 3974 is ,

predicting salary in 977. This control variable is necessary since differencesin-salary previous training may have'in effect on how much training anindividual receives. Moreover, the pre-training salarplevel 'may also, influencesalary step increases leading to the 1977 salary. By controlling pre-trainingsalary approximately in 1974 or early 1975. e should be indirectly.controllipg

q.

25

0,

-24

Table 6 : 'Effect of Schooling on Duration andExpense of Job Training; Cross-Section,,

1974-1977.

Effect of Educationby Modela Hours of Training

_.----:---,I --.

.

,; t

..

Cost ofTraining

Model 1

Model 2.

Model 3'

R2 in full model

(adjusted)

_

3.82***

3.56***

.04

.041

23.91***.

22.70***

-4.17

.029

(N = 10,360)

a: Model 1 effects are without controls. Model 2 effects are controlled forexperience and experience squared', age and age squared._ Model 3 addition-ally controls for'the effects of'occupational stream (Professional,

* Administrative,, Technical, Clericaland Other).

O

. '1

F-

26

-25-

. Table 7: Effect of Training,NEducation, andExperience on 1977 Salary for FiveCohorts

Independent Cohort, by Entry Year

Variables' 1963-65 1966-68 1069-71 1972-73 1974

.

.

Training -.02 -.29k -.03 -.03 .69***

Years of Schooling 144*** 130*** 324*** 123** 113***

Experience -.05 20.3 30.7 121* 66.40#

Experignce2

-.02 -.15* -.03 -1.07* 1.45sr-

Age -90 6.06 124 114* 27.69

e2

,Minority Male 44

.07 -

-28A

-.17

-270

-1.24

255

-1.52*

-264

-.51

-4031

',

Non-Minority Female -494e3t* -659*** -1304*** -984*** -410**

Minority Female -493* .-597*** -1402***.

-947***, -294

Administrative 385* 255 1356*** 389 472*,,14,,'

Technical -193 -493* -1097*** -1474*** -1258***

Clerical -260 -143 -728* -1188***. -1393***

Other, -599 -1642*** -1953* -2899*** -1556***

Pre-Training Salary 101*** 105*** 70*** 98*** 93***

R2

(adjusted) .920 .918, .844 .895

N 998 1262 756 4:01 691 .

X Cost of Training 132 122 4S. 250 112 168

i Hours of Training 31 31 45 39 51

-26-

for ,such pay increases and thereby not risk attributing such increases to theeffect of training. For those persons who received no training, their pre-training salary was.se equal to their 1974 salary.

What is most Striking from the results presented in Table 7 is that IJThas very little effect on employees' salary for any cohort except the mostrecent entrants to the civil service. In the 1974 cohort, foDeVINry dollarspent on IJT, an employee received approximately sixty-nine cents in increasedsalary by 1977, net of the effects on salary due to education, age, and monthsof Federal experience. While there is a negative coefficient for all fourother cohorts for training on salary, only the 1966-68 cohort has a statistic- .0fr

ally significant coefficient. For this cohort, money spent on training appar-ently had a negative effect on 1977 salary. The results suggest that for everydollar spent on IJT, an employee would make approximately twenty-nine centsless in annual salary. This cohort is the age-lump group in which largenumbers of persons were hired to fill positions in the Federal government duringthe Vietnam War expansion. It is quite possible that given the relative 'widerecruiting net cast during this period, persons with less than optimal skillswere hired into the government. Therefore, the training given to this cohortmay be in some sense for less than sat1sfactorysworers. Hence, selection fortraining carries with it a negative connotation which cannot be offset by thepositive effect from skills learned in training sessions: This is, of course;conjecture and some alternative explanation shouTd be explored. For example,competition in the 1966-68 cohort may be so great, that time away from the worksite, even for training, results in a net diskdvihtage for employees.

In general, however, the effect of training seems to be non-signifidant(when the effect of edutation, and pre- training salary are'removed), except forthe latest cohort. This suggests that training has an effect only for the newemployee. In all other cases, the positive effect from training, if there isone, is captured completely b9 the Federal government as-an employer. Thisgeneral finding is somewhat consistent with other explanations of market be-

Ahavior in the human capital model. New employeei have a considerable amount ofinformation to learn and sort for future reference in their working careers.Training courses can provide them with the tools to perform their work quicklyand efficiently. On the other hand, older employees are more likely to be morefirmly established-in their work and in their career trajectories, so that

--training programs' do not add the same increment to possible job changes, occu-, _

pational shifts, or agency moves which could then be translated into greatersalary. Therefore it would.seem, from the employee's perspective, that trainingprograms such as those measured here provide salary increments only during thefirst few pars of Federal employment. No doubt, a week away from the regularofficeicene is welcomed by many employees, but employees should not necessarilyexpect that their salary level will be affected by such training.

Of interest also from the results presented in Table 7 is the finding that,for every cohort, minoritiennA women fall farther behind white males in salaryin the space of three years or less' with only one exception (minority roles, in

'the 1969=71 cohort). That is, even though minorities and women have disparatesalaries as compared with. white males previous to receiving training, thesesalary disparities continue to increase even when training, an internal labormarket device for reallocating skills, is controlled. Giyen the possibleusefulness of training as a renied or unequal access to education previous to

c

.

-27-

job entry, we must ask the question, what is the relationship of minority/sexstatus and IJT?

Minority/Sex Status and Institutional Job Training

4ir Of special concern here is whether institutional training programs aredifferentially allocated across minority/sex groups. In the Federal civilservice, explicit emphasis has been given to training in-equal employment,opportunity efforts. According to Executiiie Order 11478 (1969), Federalagencies are to "provide the maximum feasible opportunity to employees toenhance their skills so they may perform at their highest potential and advancean accordance with their abilities." Minorities and women, who it might be,argued suffer educational disadvantages, could benefit greatly from trainingprograms. Therefore, equitable practices In the allocation of training coursesare of particular importance to equal employment opportunity policies.

The data in Table 8, however, suggest that there may yet be someinequality in the allocation of training opportunities among Federal civilservants. Using aggretate data compiled by the U.S. Civil Service Commission(1977), we find that non-minority men are favored in hours of training, percentof all employees receiving training, and average cost of training. Theyreceived on the average about 28 hours of training each during fiscal 1976,but non-minority women received only about.16 hours. Over three-fifths of allnon-minority males received some interagency training, while less than half ofnon-minority and minority females received training. Finally, the averagecost for each hour of training received wat $4.91 for non-minority males;$4.04 for minority males; $2.66 for non-minority females; and only $2.39 forminority females. In essence,, minorities and women were trained for fewerhours than non-minority males, with a smaller percentage receiving any trainingat all; what training they did receive was.of substantially lower "quality"(as measured by cost per hour) than that received by non-minority males.

r

Of..c925te, such indicators of training. inequality might be biasedby anumbeT job-related factors not explicitly controlled in the table. Forexample some occupations might simply reqUire less training than others, andthese occupations might be filled predominantly by minorities and women. More-4P-Tmer, if there is a positive relationship between training and educationbecause employers use educationlfcredentials as a proxy for trainability(Thurow, 1969); hen those employees who hdve the highest levels of educationwill have the greatest amounts of job training (Mincer', 1962).

,v0p,

To.bnswer the quettion, "Does minority7sex status affect the opportunityto receive training in the Federal civil service?" COST of training was regressed'onto AGE, jGE2, EDUCATION, EXPERIENCE, EXPERIENCE2, PRE-TRAINING SALARY, PATCO,NON-MINORITY FEMALES, MINORITY' MALES, and MINORITY FEMALES. To answer thequestion, "Does access to IJT, aswell as returns to IJT, vary by time in thecareer?," the sample was divided into two grobps of employees: (1) new entrantsto the Federal government as of June, 1974,-who were still "on-board" in June,

O

,

a

,

0 *3

-28-

Table' 8:p. Emplo e Training by Minority /Sex Group,

for A rage Hours of Training, % ReceivingTraining, and Average Cost of Training PerWorker HOur,'Fiscal Year, 197641Wm

Training Minority /Sex. Group

Non-Minority Minority Non-Minority Minority -

Males Males Females Females

Average'Hours ofTraining Per Employee 27.6 24.7 16.2 16.8

Percent of All EmployeesReceiving Training 61.9 53.1 47.0 45.6

Average Cost in DollarsPer Hour of Tratningc 4.91 4.04 2.66 2.39

aFor General Schedule and similar employees only.

bBased on number of training slots filled, rather than number ofindividual participants.

cIncludei travel, training, instruction, and supplies. Does not

include salary of trainees.

Source: Bureau of Training (1977)

0 3

30

4.*

-29-

1977 (ENTRANTS, 1974) and (2) those employees who were on-board in June,1974, regardless of their entry date, and were still in Federal employmentin June, 1977 (ON-BOARD 1974).

Presented in Table 9 are the regression coefficients for three minority/sex groups in three models predicting cost of IJT. -

The three dummy variables for minority/sex status are net of education,#ge, and square of age for the 1974 entering group of employees. The resultsin Model 1 show that all three groups received fewer funds in IJT than didnon-minority males, and that both groups of female employees received con-siderably less than non-minority males. Among the cross- section of employeesin 1974, all three minority/sex groups received an average of $100 or lessthan non-minority males in IJT during 1975 and 1976. When a control foroccupational stream is introduced'as in Model 2, differences in access to ,IJTdrbp by about 25% for the 1974 entrants, and by about one-third for the on-board group. Nonetheless, all coefficients remain negative, and five of thesix are statistically significant. When pre-trainidg salary is entered inModel 3, the coefficients for the three minority/sex groups among the, entrantsremain substantially the same as in Model 2. This is undoubtedly 6e-to thefact that for any group of entrants during one year, a control for educationand PATCO is highly correlated with GS-grade, the determinant of salary in the .Federal white-collar service. Among the on-board employees, the coefficientsfor predicting IJT for minority/sex status are somewhat reduced in Model 3.Over.time, salary trajectories would vary by minority/sex group so that eachgroup has a different pre-training salary level by the training dates analyzedhere. Therefore, pre-training salary would have a somewhat different effect .

on predicting 13T for a cross-section of Federal employees than for an enteringcohort

L

In-summary, there appears to be differential access to IJT by minority/sex group even when education and occupational stream are controlled, withminority females receiving ftWer training funds than any other group. Andalthbugh only two of the six coefficients for miVilrity males are significantlydifferent from non-minority males, all-\six minority male coefficients fortraining received,are negative. Similar to rewards, such as salary (Taylor,1979), and supervisorystatus (Kleugel, 1978), minorities and women appear toreceive less of IJT than -do non-minority males.

However, access to IJT is important only if it contributes to the well-being of an employee. Apart from the personal gratification and intellectualstimulation which training sessions may give, we need to consider whetherthe'effect of IJT on salary varies by minority/sex group. To estimate whetherthere are, in fact, different monetary returns to IJT, an earnings function foreach*minority/sex group was estimated using 1977 salary as the dependent vari--able. Cost of IJT was entered as a predictor variable in regression equationsalong with additional control variables specified in three separate models. Theresults of these analyses are presented in'Appendix B. In general,me find thatthe returns to training are greater for minorities and women, than for whitemales. These findings are somewhat contradictory withtearlier studies(Mincer, 1962), but we believe them to be representative of the types of workopportunities available to women and minorities

31

fr,

-30-

Table 9: Regression Estimates of Minority /Sex Status inPredicting Differences in Cost of Institutional'Job Training,with Non-Minority Males Among Federal'White-Coilar Employees,. 1974 -1977

Minority/Sex Groupby Rdgression Model

Employee Group .

Entrants, 1974d

On- Board, 1914d

Model laMinority Males -48.9 -80.7***

Non-Mtnority Females -140.8*** -102.3*A*

Minority Females -177.0** -122.6***

Model 2b

Minority Males -45.7 -45.2*

Non-Minority Femayes -05.7* -59.5***

Minority Females -139.2* - -62.5t*

Model 3cMinority Males -45.6 -30.4

Non-MinOrity Females . -95,3* -28.7*

Minority Females -138.9* -34.9*

\i*

a: Variables entered into the model are education, age, square of age,'each of the minority/sex variables, length of Federal service, and

square of length of service.

b: Variables entered in the model are education, age, square of,age, PATCO, ,1

the minority/sex variables, lengths of Federal service, and square of length

of service.

ci Variables entered in the Model are education, age, square of age, PATCO, °pre-training salary, the minority /sex variables, length of Federal serviceand square of length of service.

d: Levels of statistical significance are: #, p <.l; *, p<.05; **, p<.01;

and ***, p<.001. For the Entrants, R2 of model 3 = .082, (N=691);

for the On-Board, R2 of Mo4e1,3= .041,(N=10,360).

r

,

31

In summary, we find that the Federal -civil service operates as othgrinternal labor markets might,operati. We find that education is the primarypredictor of salary, in the Federal goVernment, similar to other studies.However, the effect of education on access to training is mediated by occu-pational group, an hypothesized effect. Educational credentials also seem tobe important in salary attainment in the civil service, regardlesq of race,sex, experience, or years of schodlIng of the employee. Ftnally,)the effectof IJT on increasing salary is variable by minority/sex group as is- access tosuch job training, but in cbunte'rvalling ways.

We turn now to an analysis of mobility in the Federal civil service.

MOBILITY - PROMOTJONS AND TURNPVER

The concept of intragenerational mobility has received considerableattention since the work of Blau and Duncan (1967). Ih this section we presenta model of career attainment which fits.the characteristics or important ele-ments of the Federal civil service, the btmeaucratic labor market (referred tdhere as the BLM). We examine a status attainment model, using path analyticmethods, to assess various determinants of careers, and career change. Themodel of socioeconomic achievement examined in this section (Figure 1) isbased on that which is now standard in the status attainment literature (seeespecially Kelley, 1973), wit)) modifications appropriate to the Federal BLM.

We examine two hypotheses and one objective in the following discussion.

H(7) With entry level controlled, years of schooling will bepositively associated with promotion.

H(8) With education and grade controlled, there will be anegative association 4etween promotion and age after fortyyears of age. .

0(1) What are those entry occupations which_lmobility?

--------.. ,

- Two major dimensions of socioeconomic achievement in a BLM ac salaryand occupational prestige. Salary corresponds closely with organizationalrank, though allowance, is also made for seniority. Thus the Federal General.

Schedule pay system has up to.10,seniority steps within each of its 18 hier-archiCal grades. In part the correspondence with rank reflects the commonassumption that authority 'relations will be undermined if subordinates arepaid more than their nominal superiorv(Doeringer and Piore, 1971:86). In

addition, "the average official naturally desires a mechanical fixing of tieconditions of promotion: if not of the offices, at least-of the salary-levels....from the lower, less important; and lower paid to the higherpositions" (Weber, 1968 :963). Since such desires for "orderly" careers' .

(Spilerman, 1977) are in accord with bureaucratic rationalization and withan organrealize in the salary structure of a BLM. In short, personal income has

zaiion's need for a stable, motivated work force, they are oftend(

customar ly appeared in status,attainment and related human caPiiirliodels, buta,

.

o subsequent

33

.4

'IMRE 1: SCHEMATIC mom. OF SOCIOECONOMPC ACHIEVEMENT IN THE FEDERAL CIVIL SERVICE

ENTRY

A4ENCT

FAMPOYESCHARACTERISTICS

TINE 1

AGENCY

OCCUPATION OCCUPATION

SENIORITY ,

41SALARY

D. C. D. C.

SALARY

It Combined for simpticity under the heading of employee chareeterlaIlcat are education, age. minority

group, se:, blue-collar entry port, veteran status, kinship with another Federal employee, farm .

origin, father's occupation, and lost non Federai occupation.

. a

.34

Ca

1

-33-

salary is of grpater interest withip a BLM because,it is simultaneously ameasure of individual economic reward and a formal attribute of an

.organiza-

'tional position.

In,a BLM, where a salary structure is established and administeredrationally (in the Weberian sense),,there are at least two straight-forwardmechanisms through which salary is affected by occupatioral.prestige.. First,"there are occasionally wage relationships that are dictated'by the statusof a job which management is forced to recognize" (Doeringer and Piore,1971:87). That is, social definitions of appropriate remuneratidn ma&linkoccupational prestige with salary. Second, the salary structure of a/BIM.is.typically based on elaborate methods of job evaluation and position classifi-cation (Caplow,.1954; Doeringer and Piore, 1971; Grandjean; 1980), TheFederal government's Factor Evaluation System described in greater detail lateris a case in point. These methods depend on classifiers' ratings of fops on,rather abstract dimensions such as complexity of th'e work or the level ofkuowledge required. Any bias in such ratings from the halo`of occupational 9

prestige will create an additional linkage between 'occupational standing andsalary (cf. Asch, et al., 19381. Again, status attainment models often use asocioeconomic status scale for oecupationsrather than a prestige measure (seeFeatherman and Hauser, 1976a), but the latter is theoretically more appropriatehere. In additfon,,the 'empirical stability of prestige ratings across groupsand over time (Trefinan, 1977) suggests that the application of a nationallyderived stale within a single BLM is more defensible than would be the case forsocioeconomic status. .

Accordingly, our model postulated occupational prestigeia's a determinantof salary, and occupation and salary at earlier times as influencing occupation.and salary at later times.5\The BLM variables added to the model as determinantsof rewards include senioritY, agency, and employment in the Washington, D.C.metropolitan area (the Federal," home office"), along'with a Variety of otheremployee characteristics, both biographical (education, age, sex,-ethniciti,social origi0 and bureaucratic (veterans' preferente, entry port).

/- The mod l P is estimated separately for each cohort using.ordinary, leastsqbares regOSsion, and the regression coefficients are interpreted Is indi-cating the'direct causal influences of the variables in the model on prestigeand salary (see Duncan,'1975): In Hernes' (1976) terms, we examine differencesever Iimelin the output structure (the distribution of outcomes) and.An theparameter structure (the strengths of relationships), but ast,Ume ,constant*-process structure across cohorts (namely the causal:model sumMarized in Figure 1)..

-

History and Career in the U.S. Civil Service

Demographics. A summary view of the output structurecharacterizing,eachcohort is provided in Table 10, which presents descriptive statistics for allvariables in the model. The variable labels in the'table brief)ytindicate theoperational definitions discussed more fully in Appendix C. For the two mostrecent cohorts, two time-points in'the career cycle are examined, entry andearly post-entry (i.e., 1977), while for the Great §ociety and Vietnam cohorts,the longer time-span covered permits an examination?of entry, early post-entry

-34-.

I

TABLE.1(Y: VAIIAILE MIMS MID STANDARD mualon FOR Ian COHORTS OF CAREER FEDERAL CIVIL SIIIVAMTS6

..... Cohort:

Longterm(pre-1163)

Dote of Variable: 1963 1970 1977

Variable (and Milts)

Education (years) 14.0i -- --

2.36

Entry Age (years) 3511b -- --

7.33

Entry Alls Squared 12.86

(years squared, 10005.13

Seniority (mouths) 108.013 188.46 265.36

73.81 73.29 73.58

Minority (1.0) .141 --.348

Neale (1.0) .271.444

Slue-Collar Entry (1,0) .063b

.243

Disabled Veteran (1,0) .042 --

.275

Other Veteran,. Fre- .527 -- --

formic. (1,0) .499

Defense 4 Related .402 .368 .354

Agencies (1,0) .490 .482 .478

C. Foetal Service (1.0)* .159 .160 .160

.366 .367 .367

veterans' Adadnistra- .074 .077 .077

don (1.0) .262 .266 .267

A.E.W. t Related Alm- .054 .072 .071

ties (1,0) .226 .258 .258

D.C. (1.0) .109 .098 .129

.312 .297 .335

Occupation (Siegel pre.- 47.93 44.95 49.86

tile) 10.78 10.52 10.15

Salgry (1977e. 1000 132.06 163.97 204.29

46.27 62.65 76.57

V -2214

Crest Society(1963-65)

Entry 1970 1977

13.832.26 ''4.

31.80 -. --

9.15

10.955.98

0. 72.74 149.770 13.11 13.46

.162

.369

.333 --

.472

.074 --

.262

.071 '--

.257

.403 -- --

.491

.269 .281 .258

.444 .450 f434

.360 .348 .349

.480 .477 .477

.054 .050 .053

.226 .219 .224

.071 .082 .089

.357 .275 .284

.114 .090 .127

.318 .286 .333

45.36, 47.02 48.55

10.54 10.41 9.97

101.89 131.93 176.54

36.38 53.22 63.78

4) au

Vietnam{1966-48)

Nixonoeics1969 -71)

Waterliate

(1972-73)

Cam 1,70 1977 End 1977 Entry 1977

AP

13.74 ..-- -- 14.14 -- 13.80 -2.24 2.36 2.21

30.93.

-- 29.27 -- 27.45

9.36 9.27 8.37

10.44 9.42 -- 8.23 -6.17 6.04 5.33

0 43.43 120.14 0 81.34 0 48.11

0 10.98 12.18 0 10.26 0 7.11

.203 44 .216 .221

.403 .412 .415

A .400 -- .430 -- .460 -.490 .495 .499

.065 .046 -- . .030 -

.246 .211 .170

.042 -- .043 . .032 --

.201 .204 .177

.356 -- -- .338 -- .318 --

.479 .474 .466

.379 .371 .331 .271 .253 .260 .228

.485 .483 .471 .445 ..435 .439 .420

.286 .286 .290 .211 .210' .237 .232

.452 .452 .04 .408 .407 .426 .422

.053 .055 .062 .094 .102 .102 .107

.224 .228 .241 .293 .303 .303 .309

'.080 .082 .085 .096 .109 .121 .135

.272 .275 .279 .295 .312 .327 .342

.123 .097 .119 .120 .142 .169 .136

.329 .196 .324 .325 .350 .375 .365

44.68 45.49 47.07 46.38 48.32 44.59 46.13

10.35 10.14 9.85 11.06 10.67 9.43 9.55

101.20 120.06 162.55 103.53 156.60 97.44 130.86

37.97 '49.64 60.65 42.17 60,59 36.76 48.38

-- 1023 -- 668 -- 742

a: Data are for Federal civil servants who werefull-tine. white-collar on June 30, 1977, end who entered during one of

the period. shove. Source, variable dafinitioos, and further sample restrictions are discussed in the text and to

Appendix C. Dues are reported for each variable as of the date in the coluen.boading. with standard deviations

twaediately below the respective mans. Per variables which are accused conatant.over time. statistics are reported

only once.

,b: As of 1963: entry data unavailable.

ti

vv

(1970), and mid-career (1977). Because entry characteristics are unavailablefor pre-1963 entrants, the data on longterm employees are confined to threetime-points in the middle to late career (1963, 1970, and 1977).

Observe first that each cohort displays consistent upward mility betweenthe earliest time-point and 1977, with respect to average levels of both salaryand occupational prestige. This conforms to a view of the bureaucratic labormarket as a setting for orderly careers (see Spilerman, 107). Since the datacover only those still employed in 1977, /he apparent upward mobility pay alsobe due in part to self-selection, if upwardly mobile individuals are morelikely to remain" t the Federal service: -i

Table 10 shows that shifts in the rejtative size ofFederal agencies since1963 have come about largely, but not entirely, through cohort succession ratherthan through movement between agencies by individual members of the cohorts.Within a cohort, changesin the proportional distribution across agency cate-gories tend to echo growth and decline, but only faintly. Much more strikingare the differences between cohorts. Thus, as )Ryder (1965) points out, organi-zations (like societies) are modified more by replacing individuals than bymoving them around. .Similarly, the intercohort increase in the proportion ofFederal employees working in D.C. has come about in spite of anopposite intro-cohort tendency toward more Work-assignments in the field duriog the earlypost-entry stage of the career.

In addition to substantial growth in the Federal employment of women andminorities (cf. Krislov, 1974), the table suggests an 'intercohort decline in

1 the proportions of blue-collar entrants and veterans (see Bureau of ManpowerInformation Systems, 1976; Gartaganis, 1974). However, the figures' reportedhere eiaggerate the latter trends somewhat. Recent blue-collar entrants havehad less time to make the transition-to white-collar work, and hence fewer havedone so. Veterans enjoy some statutory protection from reductions-in-force,and hence would constitute a higher proportion of those still employed in 1977for the earlier cohorts than for cohorts which have been at risk of layoff ashorter period of time. A similar process helps account for the intercohortdecline in-average age at entry, since the employment choices of younger en-trants are generally less permanent than those made,ty older entrants.,

Given the precautions taken in constructing the cohort samples, we findthat some important intercohort differences correspond to historicalinfluences on the Federal BLM. For example, the output structure Qf theNixonomics cohort is singular in several respects. Entering during a period ofdeclining Federal employment and national economic recession, the 1969-71 cohortnevertheless enjoyed the best starting salary (in constant dollars) and the .

highest average occupational prestige of the four recent cohorts. Furthermore,it achieved in just 80 months of service anaverage salary nearly ecival to thatwhich;the, cohorts preceding it required 120 to 190 months to attain. The meaneducation of the cohort, highest of the five groups, suggests that increasingunemployment in the external labor market between 1969 and 1971 permitted theFederal government to . recruit more highly trained personnel during,this period.Thus the socioeconomic success of the Nixonomics cohort qanbe partially ex-plained by normal returns on abnormally high educational qualifications. Asresults reviewed in the following subsections clearly demonstrate, however,this is by no means the whole story. A-variety of pervasive historical influ-

.40 ences have operated to differentiate the careers of the five cohorts.

-36-

Explanatory power of the model. What draws immediate attention in thenext three tables is the considerable power of the model to account for vari-ation in occupational prestige and salai-y. Whether at entry or later in thecareer, however *Ole or elaborate the prediction equations, and whateverthe cohort 114 values,at the bottom of each table show that the varianceexplained always exceeds one-fourth, usually approaches half, and oftenexceeds 80 percent. Even the simplest "human capital" - ediction equation(Equation 1), in which tht only-independent variables are education, age, andthe square of age, accounts for about 40 to 50 percent of the variance in1977 salary JO every cohort.- In this respect, the Weberian view of a BLM asa highly rationalized personnel system receives convincing:support.

The other independent variables in the model cannot be readily identifiedas either "human -capital", or "bureaucratic labor market" characteristics. Forexample, the influence of entry occupation or starting ulary on the latercareer,may be viewed either as reflecting indirect effects of human capital, oras demonstrating the extent to which initial placement determines subsequentsuccess in'a BIM. Similarly, human capital theorists such as Becker (1964)would regard seniority as>an indicator of enterprise-smific skills,or on-the-job training, but its effects might also be attributed to BLM regulations 4

designed more to promote employment stability.than to stimulate 'investments"in this type of human capital (see Doeringer and Piore, 9971).

WhereVer the line is drawn between human capital and BLM variables, theR2 values reported in the tables suggest that the explanatory importance' of BLMcharacteristics cannot be dismissed, bureaucratic rationality notwithstanding.Under the,most restrictive assumptions, with education, age, occupation, salary,senioritr, and even minority group and sex all treated as non-BIM variables,and with the explained variance already above 80 percent, the remainingvariables still add sevefal percentage points-to the explanatory power of themodel. Thus, while the Weberian view of the rationalized BLM is supported,there are apparently other influences on members' careers in the Federal govern-ment which also warrant examination.

The explanatory.power of the model is substantially uniform across cohorts,but the effect parameters show. some important intercohort differences. In the .

following subsections, we xamine each of these effects in approximately theorder in which they appelv in the tables.'ilur discussion relies on apparentpatterns of effects,,.witho t recourse to formal significance testing 10intercohort differences.

"Human capital" effects: educeim, age, and seniority. Overall, ayear of education,returns about 2 prestige points at entry, and $400 instartingsalary. Education continues to bring dirOt returns in both prestige andsalary throughout the career, though at diMiniShing rates. These results are,consistent with the predictions of human capital theory, and our earlier analy-s s, which sees education-0-4 proxy forprOductive skills valved by employers( ,g., Becker, 1964); bUt th-ty do not rule out alternative interpretationseturnss to education could also reflect mainly "credentialism,!' or simply the

beneficial effects of the "cultural capital" that formal schooling helpsdevelop, including work attitudes and'interpersonal 'tiles acceptable to one'ssuperiors (see Collins, 19791. Except in a handful of occupations, Federal lawprohibits the use of an educational credential as an exclusive qualif4cationfor employment. However, education is explicitly permitted as a substitute for

1

-37-a

4 TABLE 11 : SOCIOECONOMIC ACNIEVEKENT AMONG LONGTEM EMPLOYEES:NET METRIC REGRESSION COEFFICIENTS FOR PRE-1963 ENTRANTS TO THE U.S. CIVIL SERVICE'

1Dependent Variable: Occupation salary ($100)Date of Dependant Variables 1963 1970 1977, 1963 1970 1977

Independent Variable

Education0 2.28 0.66 0.21 4.31 3.50 1.69

A90. 1163. 0.71 - 0.35 - 0.14 3.45 - 1.70 0.10

;14 AqUii, 1963 (1000 - 0.89 0.421 0.$1 - COO** 1.771 * 0.96

Seniority6 0.00 0.00 - 0.00 0.20 - 0.021 0.00-,

Minority - 2.79 - 0.58 0.09 - 9.60 - 4.04 - 3.68 ti

Female - 4.18 - 1.50 0.12 -23.33*** -16.16*** .. 5.99

11WI-Collar, 1963 -12.33*** 2146 2.21** 3.44 . -1243m - 0.24 *

Disabl ed Veteran - 3.75 - 0.59 - 8.40 2.02 0.23ANIFF

,0.43

Other Veterans' Preference' - 1.55 - 0.43 0.39 - 0.67 1.13 1.18

Defense/Stateb

0.32 -(6.35 0.43,.

,nik.91 - 0.47, - 9.174

Postal b - 2.80 - 1.30 0.83 -15.72*** -25.08*** 9.64*

V.A.b. 1.68 - 0.49 1.18 - 6.92 T 8.03* :- 7.86*V.

N.E.W./H.U.Deb - 3.59* - 0.77 0.47 - 2.60 10.15*.

2.87

D.C., 1963 1.46 1.61 0.26 9.33* 0.79 41.87D.C., 1970 -- -'0.56 0.30

4.

- -.5 11.20*** **9.79*

D.C., 1977 --IA 0.10 9.56*.

Occ4pation, 1969 , -- .:, 0.66** 0.16** 1.41** . 0.13' 0.08

Occupation, 1970 0.63* 0.74S - 0.01. --

Occupation, 1977 -- -- -- -- -- 0.40

Salary, 1163 ($100) 0.01 0.00 0.88 0.10*

Salary, 1970 ($100) 0.00 0.99*

ConstAnt - 1.20 '12:49*** 11.053* -109.06**, -14.92 . -17.68

R2 valuesc c. 4 .

Equation 1 414 .371 .360 .314 .329 .399 v .405

Equation, 24

-- .659 .698 .418 .794 -.890

Equation 3 .371 .659 .698 .514 .794 ..890

Equation 4 .382 .661 .698 .549 .797 ,-.410, .8934

Full Equstiun .46e .666 .'02 .569 .820 .902

-

as Ungtandardised coefficients from ordinary least squares regression with all re ant independent vast010scontrolled. Two-called significance rests Sr* :reported as follows: k. 04.05: *4 p <.Ol: "*". 134.0017-Coefficfents'significanc at the .05 level but only in a' one-railed rest are Indicated by O.

b: As of the date of thedipendent. variable.cs'IM1 112 values have been adidstof for, degrees of freedom. Equation kincludes as predictors for 114;dependent

variable only education, age, and the square of age. Equation 2 adds occupation and salary aspredictors. Equation 3 further adds seniority, Ind Equation 4 inclUdes all of the foregoing plus minoritygroup and sex. The flail equation incl)Ides all relevant independent variables shown in the table and isthe equation for which the regression coefficients are repgrted.

9 (1t, (.1

N1

-38-

PIT NUM 0311R0311101 POI 06345 410 1966-66 mum 20 101 V.11.1,CIVIL UMW10111124:18111.10110 ACSIVONNIf 18 1St 66142 SOCIr17 MO MUM 008082St

110440040

Wax N0

'Sem 400 (WOO

6461444t9b

Vtatams (1046.68)

1.64914 1.016i0 4613 2.62666 3,16446 1.7406* 2.16644 0.84*** 0.4466* 6.7866* 1.47664 10444.

0.30 0.23 - 0.300 246446 0.29 -0.92 0.04 0.31 " 0.14 2.0964 -8.04 0.02

- 0.96 - 0.33 0.43 - 3.040 0.71 0.89 0.73' 0.24 - 0.11 - 1.830 0(20 - 0.16

-- 6.66 0.61 ' -- 0.2)** 0.03 .. 0.03' 0.0444 -- *44"! - 0.100

116.9180 - 2.27** .. 1421 0.13 0.4 0.26 1.111 1.23. - 1.07' .v.2. 0.02 - 0.30 4.41..

YommaAr .1 6,01696 - mg.. - 6,20 12.06a., .12.22... .13.03... - 6.114644 - 1.76646 0.20 -10.71644 - 6.71666 7.07**4

0600.001200 54an' 43.29444 2.0644 1.110 4.33 - 2.33 - 3.60 -15.22444 1.33' 6.16414 7.70' 0,30 1.61

MOW 9esr.8 - 0.70 - 0.74 0.23 - 1.31 :0.43 1.47 2.52* - 0.37 0.27 6.20 0.16 - 2.17

Otbow Vet. Prot. 1.24 - 1.296 0.32 6.200 2.10 - 3.67 . 2.1366* 1:12*. -0.40 7.00664 1.70 0.02

101004/266sob 1.350 0.37 - 0.62 6.104041 3.100 - 6.3566 1.496 - 0.33 - 0.73 7.34664 - 8.3**** 'ULU*"

60. 14681*. ., 1.14066 - 1.1366 - 1.686 7.6366 -22.57446 12.7166* - 1.11 0.54 - 1.27' 8.0766* -25.61." 11.61..*I

CA.* 11.12 - 0.05 e, 1.37 - 2.03 - 3.03 - 3.60 - 1.24 0.64 - 1.63' - 6.70 - 8.5766. .134466*

. 0

1.1.14/140.84* - 2.030 - 1.20 0.22 4.66 6.046 3.91 - 1.37 - 0.00 0.62 13.31666 3.32 2.69

1.C.. 84400 0.97 - 8414 0.16 13.9166* 1.31 -11.36** 0.61 3.0644* 6.63 10.12666 5.736 - 0,26

14C... 1170 ...1.... 0.73 1.03 -- 6.660 14.81664 -- 2.264 - 0.46 .. 1.26 - 0.02

SP' 8.C. 1077... 0.. -. 0.63 .. -- 1.026 0.71 -- .. 13.30666

01...4s114. *wry .2, 0.616** 0.1666* 1.24*** 0.08 - 0.3846 .. 6468666 0.12664 1.58666 - 0.:8 6 - 0.00

,0Occupation. 11170 -- 0.26614 .. 1.14666 0.16 --

' Osoos0,6056(346. 1,77 -- - ., 0.4444. -

Wog,. 16662 (6266)

Wart. 1976 (2106)

11.72*

.309

0.81 . 0.01 - .. 0.72466 0.01 0.00... 0.02* .. .. 1.01.6. .. .

- 1.17 16.66416 47.0666* 74.7144* 0.41 5.14 0.67

.317

.643

.. .645.

.222 .664

.40 -.Alf

*****I $4* **I.". 'Table 11.t

s *

.

.336 .323 .414 .457

.465 .782 .036

.649. .. .751 .8571

.664 .401 .716 .t87

.666'. .521 . .827 .612

40

.342 .613

.. .734

.. .735

.363 .730

.602 .744

0.56666 -- 11.93464 0.12

0.67666

0.01 0.06... 0.166..

0.00 1.10466

11217666 -37.74466 -30.66666 - 1.73

.611 .392 .656

.674 .513 .144.

.677 .. .865.

.676 .332 .845

.1114 .514 .073

.434

.863

a

TAKE 1 3: SOCIORCONONIC ACIUVINENT IN THE DLIONOMICS AND MUM% COMORTSs

NET NETNIC RECUSSION COMIC/MS 1121,1969-71 MD 1972.73 ENTRANTS TO TNEU.S. CIVIL SEIVICZa

Colt's 3 Nipoomies (169-71) Notorlato (1972-73)

Doti Voriablos °aestivation Solaro ($100) Occuoatioa Solart($100)

%to of Deo. *apt Retry 1977 Eng', 1977 1977 hum 11177

7,a4mmatAss

, Educatios . 2.47*** 0.74*** 3.22*** 3.28*** 1.79*** 0.72*** 4.38*** 2.86***

tatty Age 0.3110.i1 0.09 1.701 - 0.23 0.69* 0.22 1.17 2.38**

Eau, -Apt (100.) -0.67 -0.18 - 0.78 0.23 - 1.00* -0.36 - 0.43 - 3.67**s

loalority -- -0.02 -- 0.49*** -- . 0.04 - 0.75***

ItInority -2.23** -0.69 - 1.82 1.03 - 1.41* -0.62 - 5.31* - 1.39

Tousle 4 -3.40*** -1.50' -11.74*** -20.49*** - 2.61*** -0.92, - 8.03** - 7,99***

Illuo-Collat Entry -9.62*** 6.04*** 1.63 - 0.23 -13.71*** 9.96***' 27.32*** - 1.26

Disabled V.c.r -4.03* 0.32 - 7.23 1.33 - 2.12 -0.40 - 1.47t

9',1Lt_ .

Other Vat. Prat. -0.71 -0.60 3.345 7.31* - 1.70* -0.25 3.19 0.28

Dafessoatatob 2.76*** -0.13 5.26' -20.23*** 0.71 -1.50* - 1.91 -12.08***

foetalb -0.78 .06*** 9.78** -11.44,4* - 1.18 . -2.33*** 29.22*** -10.78***

V.A.b 0.73 2.23* - 1.77 -16.97*** 1.66 ' -0.87 - 4.23 -12,45***

N.I.V./11.11.11.b

-1.59 -1.24 13.14*** 0.79 14- 1.40 -0.79 1.42 - 0.13

D.C., Natty

D.C., 1977

t 1.33,

--

-1.32

1.50

7.17*

--

- 7.73

23.96***

0.36 0.64

-- -0.91

11.33***

--

7.93 5

712.98N,

&minimise', Wry -- 0.31m6 1.46*** 0.27 , -- 0.67*** 1.41*** 0.29!

Occupation, 1977 -- -- --0

0.31 -- -- 0.78***

,s

.94144rY, 24417 ($100) 0.03416 0.864m, 0.01 0.82.44

Comust -0.36 10.43* -93.71'44 -48.03** 7.87. 1.63 -42.841*** -78.04***

$4/41.44, . ,

i4144cum 14 .417** .401 .447 .446 .269 ___, .330 .257 .390 ,

dEquation 2 -- .391 .546 .734 .612 .329 .776

lfIquationt3 -- .590 -. .742 -- .613 -- .790

Equation 4° .439 .590 .371 .771 .274 .614 .391 .793

Pull falsettos .484 .622 .588 .806 .332 .648 .494 .808

a.b.c: See note* to Table lit.vO

41-o

T

.

-40-

required experience, with a year of college generally taken as the equivalent.of nine months of experience, and a year of graduate study equated to a yearof experience (Couturier, et al., 1979:223) . -,

A

The effects of age are also consistent with a.human capital interpretation,though again other explanations' are possible. Age at entry, a rough proxy forpotential prior experience in the labor force, brings positive socioeconomicreturns in all cohorts, especially with respect to salary. The quadraticterm, square of age, is in general negatively. associated with socioeconomicachievement,-as predicted by a hypothesis of decay or obsolescence of humancapital acquired through experience. Even more markedly than education, thedirect effects of age. appear toeoperate primarily at entry, a possible'reflection of "orderly" careers, in the sense defined by Spilerman .(1977).That is, in comparison to external labor market findings °(see Kelley, 1973),later achievement is quite .rigidly determined by prior occupatiokand salaryin the Federal government, and correspondingly less subject to the directinfluences of other employee characteristics': Still, the sample includes onlyemployees with substantially continuous Federal careers, and hence may be'biased toward orderliness.

On first inspection, the effect of seniority also seems to. conform to ahuman capital interpretation. Within each cohort, variations in senioritymeasure differences in precise dates of entry and in time lost to discontinuitiesin service. In accord with the view that seniority represents organization-specific, on-the-job learning, each month of seniority brings a net salaryincrement in the first years after entry, amounting to about $25 for the Long-term, Great Society, and Vietnam cohorts, and about $50 and $75 for theNixonomics and Watergate cohorts, respectively. However; seniority has littleor no influence on occupational prestige, and in the three earliest cohorts,in which the model includes a third point in the career cycle, the effect ofseniority on 1977 salary is entirely mediated by 1970 salary. Furthermore,preliminary analyses showed that the square of seniority is a consistently in-significant predictor of both occupation and salary in all cohorts (leading toits deletion from the model).

For these reasons, we interpret the influence of seniority on salary, notprimarily as a return to human capital, but as a result of administrativepractices that institutionalize seniority perse.as a basis of remuperation(cf. Medoff and.Abraham, 1979). The seniority steps built into white-collarFederal pay schedules become more widely spaced with longer time in grade, and.the prescribed minimum time between grade promotions 'is greater beyond theowest-grades.. These two admihistrative features of the Federal pay systemprobably account for the trivial direct effect of seniority on salary in thelater career for the Longterm, Great Society, and Vietnam cohorts. Also, thelarger returns to seniority in the two most recent cohorts support the commonopinion that promotions and step increases are now treated as more automaticthan formerly; i.e., as compensation-for time in grade rather than for exemplaryperformance.

The effects of minority group and sex. We refer to the adlltive effectsof minority group and sex, net of the other variables in the modeil, asprimarily indicative of discriminatidn, although alternattve explanations (e.g.,possible differences in previous employment history) are not altdgether ruled

4 2' .

-41-

out by the available data. _Interaction'eflects involving race andmex (cf.Featherman and Hauser, 1976a, b) are not examined here, but, will be consideredin subsequent analyses to this report.

Much of the discrepancy in socioeconomic achievement between minority andnon-minority Federal employees can be traced to occupationat entry, whereminority entrants consistently score about.2 prestige points lower. Thisdifference pro1ably represents a -cpmbination of both occupational stereotypingby race in the external labor market, and discrimination in the placement ofentrantsmithin the Federal government itself. At least since 1963, thereappears to have been little racial discrimination in occupational achievementafter entry into the Federal-service. Net:of prior occupation, the estimatedeffect of minority status is generally negative but very weak;tonly.one of theeight relevant coefficients differs significantly from zero, \_,

2.,

Saliry differences presentianother pattern. The Great Society andVietnam cohorts show essentialh no discrimination against minorities withrespect to salary at entry, nor any significant net salary differentials by1970. The expian4tion might be sought in the vigor of the civil rights movementduring the-period, in selective recruitment and retention of exceptionallyqualified or motivated minoritylemployees, jn President Kennedy's 1962 ExecutiveOrder establishing "affirmative action" in,the Federal service, or in a spillover

_ effect from economic prosperity.and'growth,in the Federal labor force. yeti'whatever the precise mix of these historical factors (cf. Burttein, 1979)-y/the apparent equality of opportunity by race experienced by employees who *4

-entered the Federal service during the 1960s is in,sharp contrast to theexperience of the Longterm- ohort, end to the salary' differentials between.

1911minority and non-minority ivil servant which are evident in cross-sectionalanalyses (e.g., Smith, 6; Taylor, 1979).: .

,

.

. ,- Unfortunately, this sanguine conclusion does not apply to the period since1970. As the 'tables show,- Minority employees' entering salaries in the two most'recent cohorts were $180 and $550 below thole of non-minority employees, netof other salary determinants, and the 1977 salaries of minorities in the Longterm,Great Society, and Vietnam cohorts fell behind those of their non-minoritycoworkers by similar amounts after 1970. The trend .is not entirely uniformand

s the salary differences mentioned are not all statistically significant, but thedirection of the trend is. clear enough. When-"benign neglect" replaced"affirmative action" as the-key to White House racial policy during the Nixon sadministration (see Morison, et al., 1977),.the impact was more than merelysemantic. ,

,,The' direct effect of sex On occupationfdeclines over the career cycle, asdoes-the effect of minority group membership, but the.sex coefficients areroughly double the. corresponding minority coefficients: Women average 3 to 5prestige points below men at entry, net of other variables land another 1.0'3points lower at the early pott-entry career stage. The lower values on theseranges-are-found in the two most recent cohqrts, perhaps suggesting an nter-cohortdecline in occupational stereotyping by sex. For the three earliercohorti, there is essentially no Ofference between men and women in 1977occupatiOnal prestige, price pricir occupation is controlled; i.e., the chancesof mobility were equal for males and females im these cohorts-after 1970, in

,part because neither glen nor women display much mobility in the later stagesOf their Federal careers. - 4

-42-

On the other hand, the effects of sex on salary are pervasive andcumulative. Controlling for other determinants of salary (including occupa-tfonal prestige), men command entering salaries $800 to $1300 higher thanthose of women in all cohorts. At each succeeding stage in the career, thegap widens by .an additional $600 to $2000, An optimist might discount the 4especially large discrepancies in the Nixonomics cohort as due to samplingerror or as a political or historicaXaberration, and if that is done, theremaining anglysesAo suggest an intercohort decline in salary discrimination,consistent with the preceding results on occupational attainment. The early1970s saw an increased political awareness of gender-related social 'issues(see Freeman, 1973), and in particular the establishment of the FederalWomen's Program within the U.S.. Civil Service Commission. It therefore wouldnot be unreasonable to infer that a period of feminist activism has had someposiiive effect on women's opportunities for socioeconomic achievement inthe Fedefal government, just as the civil rights movement may have improved.wthe career chances of minorities in the 1960s. However, the salary disparitythat remains in 19,7 still equals or exceeds the amount by which the gap Kasnarrowed since 1963, and the recent trend for minorities suggests that it willnot be narrowed further without continued. strong political pressure. Thedisparity becomes especially striking when it is recalled that the analysis isrestricted to men,and women employed full time in 1977, with substantiallycontinuous service since entry, and that the accumulated seniority variable -

acts as a control for remaining male - female, differences in career continuity.

On port of entry. Dichotomizing entrants into blue-collar and white-permits only a limited representation of'Spilerman's (197.7) argument

regarding entry ports, mobility clusters, and career lines. Yet the degree of.disaggregation described by Spilerman, who would identify as many as 10,000separate career lines, presents severe practical probleMs even with very largesamples .(Spenner and OttO, 1979). It also raises a questidn of theoreticalparsimony. The collar-color dichotomy is certainly parsimonious, and appearsa priori to be a salient criterion for distinguishing entry ports (Collins,,,1979). Although the dichotomy is crude, its influence on career success mightbe considerable.

As it turns out, Tables 12 and 13 indicate that blue-collar entry ,is nogreat handicap to long-run success in the white-collar Federal service, Torthose who do cross the barrier between blue-collar and white=collar work. Notsurprisingly, the occupational prestige of blue-collar entrants is; at entry,substantially below that of their white-collar counterparts._ Given the

, inclusion in the analysis of only those blue-collar entrants.who later achievedwhite-collar jobs, it is equally unsurprising that they display above- average

ard,mobMi;y at each subsequent stage In the' career. Net of their lowerentering prestige, the blue-collar entrants received above-average startingsalaries, especially'in the most recent cohort. However, in none of thecohorts entering the Fedefal service since 1963 is there much differencebetween the eventual salaries' of blue-collar and white-collar entrants, onceentry salary is.controlled. The net effect of blue-collar entry on latersalary is negative, but never approaches.statistical significance. Theinformation on the Longterm cohort in Table 11 suggests that those who make -'

the switch to white-collar work well along in'their careers do suffer ashort-term salary disadvantage, but even fof them long-run direct effect ofearlier blue-collar work on 1977 salary is essentially nil.

114

4

-43-

The cumulative effects of occupation and salary. Occupational prestigeat any point in the career is strongly dependent upon occupation in thepreceding stage, an effect,which is quite similar in magnitude across allcohorts. A weaker but statistically significant lagged effect, also positive,links earliest occupation with 1977 occupation in the three cohorts in whichthe lag can be tested. Thus, the chance of upward mobility in mid-career isgreater fon those who entered in higher prestige occupations, net of all othervariables in the model. ,Although this result could be an artifact of measure-ment error (see Kelley, 1973), it may also - indicate a continuing effect ofentry, port on career -'lines in a BLM.

No consistent lagged effect is evident in the determination of salary byearlier salary. As with occupation, the direct link between salaries atsuccessive points in the career is uniformly strong and positive, an apparentexpression of orderliness in the Federal career. Salary is also stronglyaffected by contemporaneous occupation, as hypothesized. Lagged effects ofoccupation on salary, and vice versa, are negligible.

Discussion on Promotions

By way.of summarizing the varied and complex findings reported here, weshould note that our results clearly document the importance of entry ports andcareer lines as structural features of the Federal labor market, and at least,suggest their importance for other BLMs as well. For example, the effects.of'employment in the Veterans Administration reflect the presence in that agencyof a ,group of seemingly disparate career lines which have in common restrictedopportunities for salary increases after entry. Similarly, the Postal Servicedisplays a distinctive pattern of earnings over the career cyCle. Or consider .

the earnings trajectory associated with a career thdt begins in the field butends with an assignment in the national Capital. Finally, tikengh-entering

::

Federal employment through a blue-collar port has little endurt g effect on thecareers of those who then shift to white-collar work,.boMetit occupational .

prestige and starting salary are powerful predictors oflubsequent achievement.In short,, an emphasison the careerlines associated with different ports. ofentry, both occupational and organizational; is no small contribution to our .

understanding of careers..

*

Still, the careeprline perspective apparently efts in its assertion that"our implicit model of education and attainment...applies much less readily tothe...civil service World.;.with formal job structures and internal promotionsystems" (Kayseri, 1973:149): Education, age, and occupation in the externallabor market are by no means irrelevant to success within the BLM, andoint

_.: generational-,influences on socioeconomic achievement-in the Tederal service do°not differ markegly from those observed in -non-BLM samples. Nor do the effectsof'minority status and sex operate exclusively through differential access toadvantageous entry ports and career lines. As just noted, starting salary andentry occupational prestige, the blue - collar -entry dichotomy, and even geo-'gra'phic location and agency are all associated in varying degrees with thecharatteristics making a given entry port better or worse for subsequent careersuccess`. The race and sex differences.in salary which remain after such

variables are controlled almost certainly reflect more than just differentialaccess to career lines,',and probably involve some,direct discriminatiOn in .

administrative decisions about'employee remunetation. .--

4,

-44-

Only one of the two hypotheses that introduced this section is stronglysupported by the results reported. With organizational rank operitionalizedin terms of salary, the determinants of salary at mid-career, net of earliersalary, can be Interpreted as determinants of promotions. Thus the consistentlysignificant net effect of education confirms H (7). The effects of the linearand quadratic. terms for age are generally of opposite sign, as would be expectedunder H (8), which proposes a decline in promotions with increasing age afterforty. However, in only one cohort (1972-73) are the effects on salary, netof earlier salary, both significant. Nor is there any clear indication thatage forty is consistently the inflection point for the curvilinear age - salaryrelationship.

I

This section has also provided some preliminary evidence on Objective (1),concerning which_entry occupations lead to subsequent mobility. Weehave seen'that blue-collar entry is no great handicap to subsequent salary mobility, 'andthat occupational prestige mobility may pe somewhat more likely for employeeswhose entry occupations are above average in prestige. We provide additionalinformation on this objective with a different analytical approach in thesection on career ladders, below.

Turnover. First, however, we direct our attention to answering efew0,0

limited questions about turnover in employment. People who leave positions ,

in the Federal government are experiencing mobility of a sort, and at the sametime they influence the mobility prospects of those who remain. Specifically, 0 .we hypothesized that,

.0 (9) . With grade controlled, years of schooling will.be pOsitivelY-associated with turnover.

H (10) Net of grade and job series, there will be no significant differ=ence-by sex in turnover.

H (11) Length of potential promotion ladder will be negatively associated.

From our Federal Careers File, we selected two subsamples of employees:thos4 entering Federal employment at any time between 1963 and 1973, and thoseentering prior to 1963, but still on-board in June, 1965. The entrant sub--''sample spans the Great Society through Watergate cohorts of our earlier-analysis,while the mid-career subsample corresponds to the long-term cohort. Becausewe are interested in turnover, we do not restrict these subsamples to employees -

still on-board in 1977 as in the'cohort analysis. However, we do require thatthe employees remain on-board for at least three years, since our concern iswith turnover among those who would be thought of as "career" civil servants.For consistency with a later analysis of the effects of job attributes onmobility? including turnover, we also restrict the subsamples to employees' whosejobs at entry (or in 1965) are amongthoSg for which attribute scores are avail-,able from the Factor Evaluation System -- over half or the Federal white collar;work force.

with turnover.

Table 14 reports, regression results preditting ;turnover from education

V

ti

448

.g

4;

TABLE 14. EFFECTS OF EDUCATION, MINORITY/SEX GROUP AND OCCUPATIONAL CATEGORY ON TERMINATION FOR ENTRANTSAND MID-CAREER CIVIL ARVANTSa

4

ENTRANTSb

MID-CAREERc

Education -.004 -.012*a

Non-Minority Female .095*** .019

Minority Male .038 -.042

Minority Female .016 -.010

Professional .082# -.004

Administrative .061 .008'1

Technical .052# .045

R2

.030 ' .067

(

aValuts in the table are metric 'regression coefficients predicting a dichotomous variable scored 0 if the' amplo e was still employed in June, 1977 and scored fif,not. Effects and net of the other variables in

the tab plus age, square of age, veteran status (two dummy variables), agency (three dummies), competitiveappoin t, P.C. assignment, and point scoreoon four job attributes (knowledge, contacts, communication,and requir OW used under the Factor Evaluation System for assigning grade level. R2 values have beenadjusted fo degrees of freedom,.

.4 ."bEmployees whqrtntered FES-scored jobs between 1963 and 1973, who were less thin 50 years of age at entry;and who remained on-boardNat least three years after entry. N *2469. *,

cEmployees who entered before 1963, who were tn FES-scored jobs in June, 1965, who were 50'years of age orless 1,p 1963, and.who remained on-board at least three years after the personnel action recording their.1965 job. N=2096.'-

.1

,

-46-

44-

minority /sex group, and occupational category; net'of our usual controls. (Gradelevel i.scontrolled indirectly by including attribute scores from FES,. multi-collinearity.between these scores and grade being so high that a diredt controlfor grade is superfluous.) The results from Table 14 do not lend much, ifany, support to-H (R) through H (11)-.---Por entrants, education is not associatedwith turnover; for mid-careerists, however, education has a weak negative, effect,contrary to H (9)

With regard tominority/sex group and turnover, we find that non-minoritywomen have nearly a ten percentage point higher turnover rate than non-minoritymen in the entraht subsample. This result contradicts our expectation underH (10), although the same comparison among the mid-careerists conforms to ourexpectation of no significant difference by sex, as do the results for minoritymen and women in both subsampleS. Thus, for non-minority women the period ofhigh transience immediately after-entry seems to last longer than th& threeyears we'have taken-as defining "career" civil servants. Later in their careers,however, their attachment to Federal employment is not significantly less thanthat of non-minority males. For both entrants and mid-careerists, minoritywomen are as firmly attached as either minority or non-minority men.

.

Testing H (11) requires an operationalization of "length of potential, pro-,motiqn ladder." After entertaining several alternatives, we used the ratherbroad grouping of occupational category (PATCO). Administrative-occupationsspan the grade hierarchy from mid-level through the top of the,supergrades,and profession 1 occupations span almost as great a range. Technical occupationsare more restr cted, and clerical occupations cover the narrowest range of all.,Under H (11)., then, thlis sequence of occupational categories should correspond

r to a ranking of turnover rates from lowest to higheet.' However, the results inTable 14 do.not support that expectation. If anything, clerical wbrkers havesomewhat lower turnover,rates than the other groups. Given our crude measureof promotion poten0a4,"-me-4eNtion against overinterpreting this finding.,

OCCUPATIONAL CAREER LADDERS

In this section we Will address four related objectives:

0 (1) What are those entry occupations which lead to subsequent mobility?

0(2) What clusters of occupations make up career patterns in that channelsof occupational the to be established?,

4 0 (3) 'What occupations provide the ability to "bridge" the major occupationalgrqups?

0 (4) What are those -Allis and Abilities-required to progress through.career lines such as those derived from. Objectives 1, 2 and_32

The first of these objectives has already;bien considered above, using the' multiple regressionapproach standard to status attainment research. Here we

r

c

6

-47-

rely mainly, on the other technique most common to mobilitY studies, cross-tabulation. The analysis examines mobility from' 1962 to 1977 for a cross-section of Federal employees on-board in 1973, drawn Yrom our Federal CareersFile4, White-collar occupations are grouped into the'official OPM categoriesused-earlier - Professional, Administrative,-Tednical, Clerical and Other(PATCO). To deteribine the extent of movement between white-collar and blue-,collar jobs, we include blue-collar workers in tbis part of our analysis,although they have generally been excluded from our other'analyses. Theblue-collar classification into Premium Ski1104,Skilled, Semi-Skilled andUnskilled occupations, which was developed but never officially adopted by OPM,is used here as the best available classification -for our_purimes,

To simplify the discussion, we present not the raw mobility matrix itself,but the standardized lambda parameters from a log-linear analysis of thetable using Goodman's ECTA. These effects take 4into account the fact thathe categories differ in size. Positive values 4ndicate cells that include amoverrepresentation of employees, while negative-values indicate an.underrepre-sentation. Standardized values greater than 2 in absolute magnitude indicatethat the over\ or underrepresentation'is statistically significant beyond ;he.05 level.

Table 15 reports the results. The most striking finding is the concentra-.wtion of large pos,itive effects 0 the mainCilatonal. Most Federal eMplOyees,

do not change occupational categories. Indeed, among those on-board in both1969 and Q=1.9..fewer than gO,per6ent were indifferent categories in the two .,years. To pe sit this 'understates the 'total_ mobility. Some employees changedcategories more than &ice, trid a,few of these had returned,to,their,1969 cate-gorieS by 1977' F.urther, Ouctategari 'e,quite brO d, anti hence mask mobi-1 between d tailed occupations,,wftWn ttjor.categbri, s. Foy the full 1973dross-section, 29' percent hadtatlWastln han4ei9f :det'iledwoccupation be: ,

tween 1973 an 1977, and 9 percenlOad at let%t;two,sbch hanges over thoseyears. Finally, there is a substa tial amount. of mopility;out of the Federalikovernment -altogether. Over a qua er of those'4mgked in bott 1969. and 1973

left the government by 1977, w le nearly hal!, Of those entering between i9 and 1973 were goneby 1977. Still, the Ob ious.conclsiOn from Table 15

is that most Federal careers are confined w4hi broad occupational categories.

Occupational ftslhat do occur take pl ce, largely within the blue -.collar /white-collars oupings indicated by_the dashed lines in Table 15.Among the white-collar occupaNions, for example,*the professional and adminis-'trator categqries feed each other employees, While the technical workers transferinto both the professional and administrativeicaregories. Clerical workers, onthe other hand; appear barred from entry into, the professiignal occupations, butare able to move into both the administrative-and technical ,groups. In no casewere white-Collar workers likely to move intO anxdf the,blue-collar occupationalgroups.

Among the blue - collar worrs, there is'a greater amount of what mightber' considered downward movement than among white- collar, workers. For example,premium skilled workers -ire fairly likely to:move.into.the lower category ofblue - coll=ar work (skilled workers), and the Skilled workers themselves show a

Z.

6.

Table 15: A Log=Linear Analysis of Occupational Mobility, 1969 -1977:

Standardized LambdAa

Occupational Group, 1977

Occupational Professional Administrative Technical*

Clerical Other PremiumGroup, 1969 Skilled

Skilled

SkilledSemi- Unskilled Leavers

________ __Professional---143.35 6.73 - 2.11 - 2.67 0.87 I - 0.87 - 2.29 - 1.62 - 1,04 5.81

Administrative 2.23 22.48 0.80 2.48 - 1.04 I - 0.87, - 3.39 -2.70 - 2.11 I 4.60

Technical 3.15 6.69 21.04 0.99 - 2.43 I - 1.52 - 3.72 - 2.62 - 0.15 # 0.624. .

Clerical - 5.47 4.23 8.95 30..08 - 1.88 ..- 3.32 " 4.82 - 1.64 -.2.35 4.19

Other 2.18.

.1.874 -2.85 - :- 3.32 16.80 - 0.82 - 2.73 -60.66 - 0.65 - 5.87

Premium Skilled - 2.10 - 2.56 2.78 2.65bw - 0.09 I 14.94 6.11 - 1.70 - 1.11 - 1.53

Skilled 7 - 6.89 - 9.60 - 6 17" 7.15 ; 2.29°1'°I 7.17 17.71 5.72 2.17 - 7%14

Semi-Skilled - 4.89 - 7.08 - 1.8 1.93 -1.71 - 0.17 7.66 16.56' -2.24 -'2.13

Unskilled -- 2.94 - 5.31.4s. '-'4.34 0.01 0.87 1.51 3.87 6.64 14.64 -1;.54LEntrants 3.72 - 7.41 - 3.42 10.35 1.26 I - 0.89 -. - 0.91 0.60 - 0.05 - 8.73

a: N28,690

"'1111111111111

p.

. ,r"

S .

large chance Ompovement not only into the premium skilled group, but alsdinto the semi-skilled and unskilled groups. Simtlar is the situation for thesemi-skilled and unskilled groups.

In the main, the patterns of mobility displayed in the 1969-77 table arestMilar to those apparent over tile four-year periods, 1969-73 and 1973-77.We therefore omit-these results here.- The three-way table, 1969-73=77; in-cludes 900 cells and thus defies parsimoniOus discussion. Since many of thecells are empty, signifying rare mobility histories, log-linear analysis isproblematic. To indicate in general terms the:kinds of mobility patterns '''

revealed in the three-way tables, we present (in Table 16) the 1973-77 mobilityof employees who in 1969 were in the largest major category, the clerical,w*ers. (We are grateful tofty Aldred for compiling this table.)

The table shows, first, that there is little return mobility. That is,only 1 percent of employees who were clericals in both 1969 and 1977 were admin-istrators-or technicians in the interim. Sec vd, it shows that thervare tworoutes from clerical intO'administrative jobs. Most who make that shift appearto do so directly, whether between 1969 and 1973 or between 1973 and 1977.Still, nearly 10 percent Occupied technical jobs as a "bridge" between their.1969 clerical and ,i977 administrative positions.

Because of the relatively small numbers of employees who change majoroccupational categories; direct identification of the detailed occupationswhich might serve as bridges between categories is not feasible.- Instead,our analysis of Objectiye (3) focuses on the job attributes that appear to

'promote such bridging. In turn, the analysis of job attributes providesclues to the skills and abilities associated with mobility in the Federalservice=-our Objective (4). That is, by determining the attribOtes of jobsthat are associated with occupational chafte, we hope to shed light, albeitindirectly,on-the types-of abilities that education for career preparationmight seek to develop.

The job attribute.measureS used are taken from the Federal government'snew Factor Evaluation#,Ustem (FES) of position classification. When FES isfully implemented, 6tle'r&ccupation in the General Schedule pay system willhave grade levels aSt490:4 according to scores on the nine job factors definedin Table 17. These nine factors are: (1) Knowledge required-by the job;(2) Supervisory control;,(3) Guidelinq; (4) Complexity; (5). Scope and °effect;(6) Contadts; (7) Purpose; (8) Physical requirements;_and (9) Work environment.We have obtained the factor scores for nearly 500 occupation/grade combina-tions, which include over-half Of all 'General Schedule employees. We reliedon official published scores whenever possible, but also used, where necessary,draft scores still under review and trial scores from early field tests of°FES. Where more than one score is published for, a given'factor on a singleoccupation/grade combination (in FES jargon, when a job has more than one"benchmark" or typical combination of duties), we,took the simple arithmeticaverage as our best approximation.

We-report separate analyses for, the two subsamples whose! turnover Was 4

exa

51

.

-50-

Table 16: Selected 1973 and 1677 Occupational Categories of1969 Clerical EmOloyees (Percents)

1977

,

1973 A T C L Total

A 41.2 1.7 .4 1.7 (220)0

0.

T 9.7 35.6 .6 3.8 . 'i. (314)i e

C 49.0 - 62.7 99.0 94.5 (7236)

Total 99.9 100.0 100.0 100.0 (8166)

(383)- (464) (5276) (2043)

52 .

4

1

4.

-51-

Ttita.5- 17 FACIPRS IN THE FACTOR EVALUATION SYSTEPa

Factor/Sukfactor Definition

Knowledge Required

Responsibility

[Lack of) Supervisory The'nature and extent of direct or indirectControls controls exercised by the supervisor, the

employee'eresponsibility, and the review ofcompleted work.

(Lack of) Guidelinagi The nature of guidelines and the judgmentneeded fb apply these guidelines. Jobs varyin the specificity, applicability, and avail-ability of guidelines for performance ofassignments.

The nature and extent of,iaformation orfacts which the worker must understand todo acceptable work (e.g., steps, procedures,practices, rules, policies, theory, princi-ples, and concepts) and the.hature and extent

of skills/abilities necessary to apply theseknowledges.

4Difficulty

Complexity The nature, variety, and intricacy.ofthework performed; the breadth, depth, or extentof tasks or projects assigned; the'relatedor conflicting information, programs, orconcepts. -NA

Scope and Effect The purpose of ssignments; the nature ofthe cases, questions, and probledm involved:andthe effect of the completed work bothwithin and outside of the organization.

Personal Relationships',Personal Contacts

Puroo;e (of Chpiacts)

Environmental DemandsPhysical 'Requirements

Ranges from conticts.witn other employees inthe immediate work unit ... to contacts withhigh-ranking officials outside the agency.

[Ranges from contacts) to obtain, clarify,or give facts or information directly relatedto the work (to contacts) to justify, defend,negotiate, or settle matters involvingmig-nificant or controversial issues.

.11

'the requirements and physical demandi placedOn the employ 'by the work assighment. .

This include physical characteristics andabilities .g., speCific eyesight and dex-terity req irements) and tfle:physicil exertioninvolved in the work.

WOrk Environment The risRs, discomforts or unplei'santness thatmay be imposed upon employees ly variousphysical Burr ridings or job situations.

aSource: Verbatim, except for,bracketed inserts, from Personhel'Research andDevelopment Center (1973:77,79,42,84.,86,88,90-92).

53

examined earlier: entrants between 1963 and 1973, and mid-careerists as of1965. The first'part of the analysis involves the computation of. what weterm "continuity coefficients." These relate the attributes ofan empfoyee's

job at entry (or in 1965 for the mid- careerists) to the same attribute in1977 or atftermination. The size of the correlation thus gives the degree ofab continuity in that attribute. The attribUtes were scored in two ways forthis analysis. The'"raw" scores are simply the nine FES factor scores for-

a the'employee's occupation and grade.' In addition, we have generated whatwe call "profile" scores, obtained by (1) converting all nine of the factorsto a common metric, using Z-scores plus the constant 10 to insure pos.itive-values'; (2) summing the nine factors in the new metric; and (3) expressingeach factor in the new metric as a percentage of the total score on all nine.The resulting profile scores give the relative predominance of any one factorcompared with all the others', rather, than its bsolute magnitude, and thusare more informative for some purposes. In Articular, they reduce the auto-correlation between attributes at successive times. Also, multicollinearitybetween the profile scores and GSLgrade is less-pronounced than is true ofthe raw scores.

Table 18 shows a high degree of continuity on all attribiltes for theaverage Federal employee over the early career, though typically,More so '

for the raw scores than far the'profiles. Pairs of raw scores at successivetimes share from half to three-quarters of their variance; pairs of profilesshare a quarter to half. Similar conclusions hold when we look only atemployees who to any way change jobs (either occupation or-grade, or both).However, about half of the latterjroup did in fact change occupations whenthey changed jobs, and for these employees the continuity coefficients areconsiderably less. The4raw scores share only a tenth to a quarter of theirvariance; and the profiles at most a tenth. Using the'profiles as a guide,the continuity between entry job and :later job, for those who-change occupa-tions, is greatest with respect to knowledge, responsibility,(absence of?supervision), personal contacts, physical, demands (including manual skills);and environmental conditions. If we loorkvat employees who not only change ,

their detailed occupational series, but also move,between major occupationalgroups, we see even less continuity. Again using the profile scores, onlyknowledge and physical ,demands yield significant correlation.

In Table 19 the analysis is repeated for the mid - careerists.. The patternof findings is generally similar, but there is substantially greater-continuity.Thus mobility in the early career appears more random than later in the "career,,in that it tends to-fWvolve less similar occupations.

Among the entrants, especially, the attributes that display the gregtes.t. continuity are knowledge and physical requirements. This is a finding of no ,

small import, for.ittis precisely these two attributes which seem most directlyrelated to trainable abilities of individuals--general command of a subject-

$ matter, on the one hand, And specific technical skills, on the other.

The foregoing results speak to the question of what attributes an originand destination, job tend to have iwcommon, given that mobility occurs.

51

8

/a

e

t

. '1Table 18

Continuity Coefficients for Ram Scores and Profile Scores on Nine Attributes, by Nobility:. Entrants, to the U.S. Civil Service, 1963-1973°

Al l- Employeesb

Attribute , Raw. !Yoffie

.872 ., .683

.702 :509 de

.760 .57',

.794 .565

.809 .582

8.Contacts .731 .555

Communication .714' .480

Requirements .771 .781.......

Environment .766 .768

2262 2262

Knowl edge

N Supervisiory

Guidelines

Coippl exity.

Scope

.sr

e

N

0..

1Job ChangerscRaw Profile

.852 .595

.658 .417

.714 .435

.76T ,.430

.786 .462

.680 .438

.654 .332

.710 .729

.724 .718

1834 :1834

i

Series ChangersdGr'up Changers e

Category ChangersRaw Profile Raw Profile Raw Profile

..533 .112 .503 .166 ..410 .169".368 .175 .339 .082n .377, .180**

.387 .067* .358 .038n .375 .100n

.396 -.016n .389 -.027n .284 .091Nej

.48' .096** .399 .037n .357 .078n '

.346 .130 .130** -.017n .347 .170**

.304 .077** .272 -.018n ,.386 .00.281 .310 .105* ,.188. .'266 .323

..318 .221 .036n .038n_ .320 .323

923 923 367 . 367 202 202

a

a: Values in the table are Pearson (product-rnoment)'correlation coefficients between'each job attribute at entry and the same attribute as ofJune, 1977 (or at termination,). All coefficients are significant at the .001 level (one-tailed test), expept as follows: **, .01; *, .05;n, not significant. fb: Employees who entered between 3 96p and 1973 and who were in sco'red jobs at.entry and in June, 1977 (or at termination), 'Employees whotennina'ted less than three years after the personnel action recording their entry job or older than 50 at entry are excluded.. .,..

.

c: A subsettime points.

d:.-* A subset

e: A subset

f: A subset

55

of'the employees described in note b, who changed either gractejevel or occupational series (four-digit code) between the two

of the employees described in note c, who changed occupational series.

of the ernployees.descrited in note d, who changedof.the employees described in note c, who changed

A

...

occupational group (two digit code).occupational category (one-digit PATCO code).

8

5G ,

;

.

Q

Table 13.

Continuity Coefficients for Raw Scores and Profile Scores on Nine Job Attributes, by Mobility:Nicf-Career U.S. Civil Servants, 1965a

Attribute

Knowledge

,Supervision

Guidelines

Complexity

Scope

Contacts

Communication

Requirements

Envirimment

N

All Employees bJob. Changersc Series Chhrigersd Group ChangerseRaw Profile Raw Profile Raw Profile Raw Profile

.900 .761 .874 .617 .751 .311 .694 .338.779 .598 .708 .406 .558 .116 ** _499 .042n.830 .649 .754 .426 .682 .184 .610 .142*.832 .682 .769 .460 .682 .175 .648 .186**.846 .664 . 02 .447* .720 .343 .671 .283.736 .594 .5% s.368 .475 .166 . .362 .070n.774 .599 .667 .340 .499 .095* .448 .004n.851 .872 .762 .814 ,403 .520 .136* .437

.841 .848 .740 .764 .309 .349 - .074n .183**1729 1729 1164 11 64 511 511 218 218

Category ChangersRaw Profile

.52Z. .221**

70 .209".

*....-,.451 92*

.516 .221**4-',

.526 .309',2

.287 .147*

.292 .bo7h

.605 .560

.272 .1 50

172 172

a: Values in the table are Pearson (product-moment) Correlation coefficients between each job attribute in 1 965 and the same attribute as of June1 977 (or at termination). All coefficients are significant at the .001 level (one- tailed test), except as follows: *1'; .01; *; .05; n,not significant,

b: Employees who entered before 1 963 and who were in scored jobs in June, 1965, hnd in June, 1 977 (or at termination). Employees whoterminated less than three years after the personnel.action recycling their 1965 job or older "thilh 50 in 1963 are excluded.c: A sutiset of the employees described intine points:

d:

e:

f:

note b, who changed either grade

A subset of the employees., escribed in note c, who changed occupational

A subset of the employees described in note d, who changed occupational

A subset of the employees deScribed in note c, who chAiged occupational

57

level or occupational series (four-digit code) between the two

series.

group (two-digit code).

category (one-digit PATCO code);;

53

1F. -55-

Another way to approach the issue of skill transferability.using the jobattribute data available to us is to ask what attributes of an origin jobare most conducive to subsequent mobility.

Results bearing oa-this question are reported%in Tables 20 and 21 forthe entrant and mid-career subsamples, respectively. Takdng each of ninedifferent kinds of mobility as the dependent variable in a regressionequation, we examine the effects of four of our job attributes; net of theusual controls. Nlticollinearity prevented us from simultaneously examin-ing the effects of all nine FES factors, so we selected the two with the'greatest continuities in the previous analysis -- knowledge and physicalrequirements -- plus two that seem especially likely to improve an employee'schances for mobility -- contacts and communications with a network of othergovernment and non-government workers in-the course of the job. For simplicity,

' we use ordinary least squares regression techniques to estimate the effectsalthough there are more sophisticated methods available for dichotomousdependent variables.

The tables indicate that noneliof the four job attributes we selectedfor study has much effect on most kinds of mobility. Looking first at theeffects of the raw scores, we see that employees scoring higher, on know-ledge, communication, and to some extent-physical requirements are fess likelyto change grades. This may be because gre0e promotions (by far the mostcommon grade change).-are less frequent at higher levels. Those scoring highon knowledge are also somewhat less likely'than others to have a temporarybreak in their Federal employment. The last result bolds for the profilescores as well, but the negative effects of the raw scores on grade changebecome positive when the profile scores are used. This suggests that jobs'which are relatively high or knowledge or physical requirements do providetheir incumbents with transferable skills that canbe used to climb thegrade hierarchy, but.that this effect is, masked in the raw scUre'analysisby the effect of grade per se. For none of the other types of mobilityexamined are there strong and consistent effects, either positive ornegative, from the job attributes, considered in the tables.

Summary and Conclusions

Hypotheses and objectives. Where previous theory and resew h allowedus to assert specific, determinant hypotheses in our original proposal toNIE,,our results can be summarized simply by reiterating the degree of supportwe have found for each hypothesis in our Federal careers data'. The readershould of course refer to the body of our report for qualifications andreservations concerning the findings so baldly stated here. Thus:

H(1). Years of schooling and amount of experience willbe positively-associated with Salary. Supported.

H(2Y. The associations in H(1) will be stronger for whitemales than for non-white males and feMales. Ambi uoLlt-

supported in cross-sectional but not in cohort analy

H(3). The highest returns to schooling will be for th'o'se'

5s

TABLE 20. EFFECTS OF SELEC1ED JOB ATTRIBUTES (RAWSCORES AND PROFILE SCORES) ON MOBILITY:

ENTRANTS TO THE U.S. CIVIL SERVICE, 1963-1973a

sa

Dependent Variable,44san)

i 1

Raw ScoresProfile Scores

Know-

ledgeCon-

tactsCommuni-

cationRequire-ments R2

Know-

ledgeCon-

tacts .Communi-cation

RequIre-ments R2 4'...

Geographic Move Li.448) .008 .072* -.048 -.029 .109 -.020 .029 -.035 -.0461 .109Agency Shift(.261) -.071 -.001 .007 -.053* .124 -.005 .001 -:603 -,039 .123Job Change(.886) -.278*** .013 -.171**k -.0451 .137 .078. .049* -:065** .096*** .111Grade Change(.856) -.354*** .010 -.169*** -.053* .143 .099** .065** -.049* .128*** .110Series Change(.572) -.041 -.033 .008 -.049* .295 -.003 -.022 -.004 -.040 .294Group Change - (.260) -.030 7.024 -.029 -.021 .105 .045 .005 -.005 .022 .104Category Change(.193) -.031 .012 .039 -.023 .133 -,018 .004 %017 -.029 '.133Break in Service(.282) -.191*** .020 -.027 .050* .149 -.095** -.001 -.000 .0511 .149Termination(.148) -.015- -.002 -.0600 .056* .030 ..038 4%007 -.012 .083**, . .022

aValues in the table are standardized partial regression coefficients predicting the iaicated dependent variable fromeach of four attributes of the employee's entry jobs.net of the other three attributes an8 the following controlvariables: education, age, square of age, veteran status (two dumMy variables), minority/sex group (three dummies),agency (three dummies), occupational category (threedummies, competitive appointment, and D.C. assignment. Forall dependent variables except termination, mouths at risk of mobility (entry date to termination date) is also con-trolled. Dependent variablei are dichotomies scored 1 if the employee experienced a given type of mobility at leastonce betweenAitry and June, 1977. or termination): scored.° otherwise. See notes t6 Table 4- for definitions ofjob, series, group, and category changes. 'Significance levels are reported as follows: ***, p<.001; **, p< .01;*, p<.05; I, p 4.05 one-tailed. R? values have been adjusted for degrees of freedom. N = 2469 employees enteringa scored job between_1963 and 1973. Employees who terminated less than three years after the personnel action record-their entry job are excluded, ac are those older than 50 at entry.

cp

4.

rI.

'TABLE 21. EFFECTS OF SELECTED JOB ATTRIBUTES (RAW SCORES AND PROFILE SCORES) ON 'MOBILITY:MID-CAREER U.S. CIVIL SERANTS, 1965a

le

a-

.4. ,

Dependent Variable (Mean)

Raw Scores Profile Scores

Know-

ledge

Con=tacts

Communi-% Requfte-cation. ments .R2

Know-

ledgeCon-

tactsCommuni- Require -

cation R2

Geographic Move- --

AgenCy Shift

Job Change ,

'Grade Change

Series Change

Group Change

Category Changes

Break in Service

Terminaiion ,

(.469)

- (.297)

(.800)

(.742k

(.474)

(.222)

(.216)

(.312)

(.209)

-.081 ".

-.074

-.290***

-.303***

-.004

-.001

.030.

-r135*

.029

-.014

-.033

-.035

-.024

-.071*

-.039

-.009

-.040

-.062

.024

.015

-.066*

-,078*

-.077**

-.066*

4

57#

.073*

-.002

-.090***

-%029

-.0&8

-.093***.

-.082***.

-.093***

.024

-.019

.125

'.158

.236

.228

:267

.098

..131

.173

.067

.012,

.036

.00f

.015

.091**

.076*

.106**

-.003

.017

.007

-.007'

-.006

.010

-.027

-.004°

.031

-.003

-.026

tments,

.028 .034

.420 -.048

-.036#

-.037#

.077*

.099**

- - -.038# -.031

-..030 -.003

.014 -.041

.050* .,084 * *,

.055* -.008

.124

.155

.217

.209

.261

:095

.133

.173

.067

0a

aSee liote'to Table 20. N=2096 employees whd entered before 1963, who were in scored jobs in Julie. 1965, who wereyears of age or less in 163, and who remained on-board at least three years after the personnel action record-

' ing their 1965 job.

ti

3,*

O (

employees who have high levels of education andhigh levels of.experience. Supported.

H(4): Returns.to schooling and to experience will besimilar in magnitude. Not supported; returnsto schooling greater.

H(5). Conteolling for years of schooling and governmentexperience, earned degrees will be positively

, associated with salary. Supported.

H(6). Duration and expense of job training will be posi-tively associated with years of schooling. Supported,but relationship mediated by occupational stream(PATCO).

H(7). Withrentry level controlled years of schoolingwill be positively associated with promotion.Supported.

a`

H(8). With education and grade controlled, there will bea negative association between promotion and age

y.after fortyears of age. Weakly supported.

HcB.). With grade controlled, years of 'schooling will be' positively associated with turnover. Not supported.

10). Net of grade and job series, there will be no

signifidant difference by sex in turnover. Support-ed for minority employees only.

H(11)% Length of potential promotion ladder Will be nega-tively associated iiith tufnover. Not supported.'

,p

... .10n the other hand,'where our research objectives could only be stated.'as questions, not as specific hypotheses, our results are more difficult to

ityrsummarize. As we noted in our proposal to NIE,,. 'lacking the well-integrated

-

foundation of previous theoretical and empirica :'. rk which characterizes theliterature on human capital and social mobility, t e career' development por-tion of our research will be mainly exploratory." One of the more general

, research-questions, dealing with the effects of college, major on careers, hadto be abandbped altogether because of excessive missing data On the key variable,college major. The remaining qUestions address the infFPelationships among

4-'it,c,0"ers, occupational Change, and transferable skills, and here we regard ourresults as suggestive but by no means definitive. Again quoting from our grant-pr osal, "our over-riding objective throughout this...area of the research[has been] to raise questions no, less than to find answers."

Briefly, then, what are the tentative conclusions that we draw fromthis phase of the project? First, we share the growing conviction that the .

term-career should be defined quite simply aS" an individual's sequence of jobs

C2

a

over time (cf. Spenner and Oho, 1979). The extent and form of regularity,orderliness, or stability in careers then become topics for investigationrather than matters of definition,,as in the traditional usage:

Second, we find that careers in the U:S.civil service are characterizedby a high degree of orderliness, in Spilerman's (1977). sense of regdiar advance-ment. .The-amount of mobility between detailed occupations is a4o4t as high asin the-general labor force (see Sommersand Eck,, 1977), but there is relatively

.

little movement across major categories (PATCO). Again like the,general labor-force (see Blau and Duncan, 1967),here is some movement from ,) blue-collar-to white-collar,'but very little in the reverse direction.

\4'.__.)Third, patterng of occupational change, and career'success more generally,

*ear §trongly influenced by the structural characteristics of the Federalgovernment as an internal labor market. We have not isola ed specific detailed

;occupations which are consistently the most advantageous p its of entry or whichserve to bridge major occupational categories. However,we pave shown the lastinginfluence that placement at entry can !lave on the career and have shown furtherthat the relevant characteristics ofan entry job are not just its occupationaltitle, but also the geographical and departmental locations of the position with-in the organizatti.

,

.

f ., J

Finally, we have sought to add.ess the key question posed in the NIEgrant announcement under which our project was funded', "To-what extent may a

11, series of jobs held by individuals be related by similaritiesof skills, abilitiesor attributes required by the job?" Our answer to this question depends. onwhether the concern is for the minority of. employees who Make.major occupationalchanges or for the majority who rarely if ever do so.

If the latter,, the degree of continuity in job attributes over time isvery high. That result hold whether we consider recent entrants Or employeesin the middle of their careers, and whether we use the ,absolute level of eachjob attribute or a "profile score" giving an,attribute's relative preponderancein a job's total tasks.,

. y.

However, examining major occupational change, we find much less evi-dence of attribute continuity. Especially for the profile scores of recententrants, the attributes of jobs tend not to be strongly Telated'over timeamong employees making a major occupatiOTial shift. Nor are the attributesvery good predictors of the likelihood of sdth a shift. -

In short, to the extent that skijrtransferability can be inferred 6from attribute continuity, such transfer has probably not played a big partin facilitating occupational- change for Federal civil servants. Most employees

,.

make no major Occupational changes; their skills "tpansfer" because their jobsdo,not.change much. Those who make bigger changes show no strong tendencyto move into jobs with similar attributes, where their skills might seem mostreadily transferred. Indeed,.some fraction of them may move in part because %

7' the ,are better suited by skill or temperament to a\job with very differentattributes.

(e3

-60-

The job attributes with t14- highest continuity and, by inference,the skills with.the,greate9t transferability -- pertain-to the level of +now-ledge required and the physical demands of the job-, including specific technicalabilities. This'resUlt runs counter to much of the theoretical work on trans- /ferability, which argues that more general competencies such as, dealing withpeople are more transferable (e.g., Freidman, et al., 1978). We can suggesttwo possible interpretations for further research to addresq. First, it maybe precisely the mastery of a specific subject matter or technical skill --say, tax law,or drafting -- that provides the bridge between otherwise dis-similar occupations. For example, a tax lawyer may become an administratorof a tax-related program. Alternatively, it may be that the skill actuallybeing transferred is the-ability to learn per se. That is, the ability tomaster a given subject matter or technical skill may itself be a skill thatimproves with practice, so that even if the content of.the knowledge requiredin a new job is different,,a facility at acquiring such knowledge may transfer.

While the mechanism admits of at least these twosinterpretations, theimplications of our finding for educational practice are much the same in eithercase. A concern for teaching-transferable skills to prepare students for acareer of occupational change should not sacrifice thorough mastery of specificsubject matters.

`Generalizability. One final question demands attentin before we conclude .;'

is,report: to what extent can our study of careers in the Federal civil serviceeneralized to other employment contexts? Although not explicitly formulated

as a research-objective, that question has been a matter of concern fromitheoutset.

i

Career research has only- recently begun to specify the effects oforganizationaj context 'on the socioeconomic career. The notion of an internallabor market, discussed in detail by Doefinger and Piore (1971), draws attentionto the effects of formal rules and informal customs on organizational careers.Other' recent work (see Althauser and Kalleberg, 1977; Beck, et 21., 1979; Formand Huber, 1976) documents how diffefent labor markets affect people's workhistories in very different ways. Theory and some evidence also suggest thatone's location on the various dimerisions along which an organization is differen-tiated can have conseqUencesfor career success: the vertical differentiationof authority (Kluegel, 1977), the horizontal line/staff distinction (Pfeffer,1977), the geographical distribution of an organization's activities (Talbertand Bose, 1977), and differentiation into separate units or departments (Martin

,and Strauss, 1959)". Opportunities for advancement may be greater in the homeoffice, but some experience in the field may be essential preparation for thehighest positions in an organization. If the experience is in the largest

).plant or dppartment, or the one most closely identified with the organization'smajor product, so much the better.

tOrganizational labor markets are a heterogeneous lot: Consequently,

the effects of organizational context may go unnoticed in a standard humericapital or status attainment study, Even one which explicitly includes organiza-tional variables. Such studies typicallY'lump workers from diverse organizationsin a single analysis, averaging away the effects of context Recent career

S)

C-I

-61-

research has therefore come to recognize the need for longitudinal case studiesof specific organizations,'to isolate organi2ational influences on careersuccess (Kalleberg and SOrensen,. 1979),

The project reported here is an application of that strategy; our case,

the U.S. civil ser ce. As the nation's largest employer, the Federal. civilianlabor market is w rthy of study in its own right, both because of its directimpact on the nearly three million workers it employs andlor its indirect ,effects,

,

through policy formation.and implementation, on the lives of millions moreButBut civil service careers-are also important for what they can reveal, and forwhat they might suggest, about organizational careers more generallx..

Depending on how one counts, from a quarter (Stinchcombe, 1965), tohalf (Caplow, 1954), to three-quarters (,Doeringer and Piore, 1971) of employedAmericans.work within bureaucratic labor markets. Nearly sixteen percent workfor government at all levels, another quarter of the labor force work forbusinesses with more than 500 employees, and still others are employed bylarge private universities; hospitals, and the like {Kanter, 1977:15). Of

course, the Federal civil service is hardly "representative" of all thesedivei-se organiiations. Its size alone makes it atypical, and the absence of aprofit orientation is certainly significant. .Perhaps most important, Federalpersonnel regulations have the force Of law, and some of these (e.g.., theveterans' preference) are peculiar to public employment. But our case study

.does permit us to isolate the kindstof mechanisms that determine career successin, such labor markets, even if the,Oecifics.are.likely to be more idiosyncratic.

Generalizability.is probably greatest from the Federal civil serviceto the centrar.administrations of other countries, and to State and ;local .

governments in the United States. From a review of_the available cross=nationalevidence on ci5vil' service careers, Sheriff (1976:54) concludes that "thesimilarity...across such a diversity of- systems is fairly astonishing...It isalmost as if there were is. 'Weberian.,myth' of how civil service organizations

.\ ought to be run." Simple cultural diffusion explains part oflthe similarity,since the U.S. has often'served as a model for other countries, includingeven some aspects of public administratidn in the Soviet Union and Eastern Europe(see Cayer, 1975:147). In turn, the civil service reform, movement in Americajythe late nineteenth century, which established the modern civil service,gas ,heavily influenced by the British and Prussian examples (Shafritz, 1975).

By the same tt8ken, most states and the larger lobal juilisdictions inthe U.S. have "merit" employment plans niodeled after the Federal personnel

system. Initially, the imitation was sincere flattelly, but in recent decades

it 'has been in pdrt coerced: Federal statutes increasingly require meritemployment practices by state and local agencies handling Federally-fundedprograms (see Meyer and Brown, 1977).

Imitation and coercion explain some resemblances between the .ederalcivil service and the personnel practices of large Orivite businesses,,,too.For instance; examinations.for job applicants are now among the standard toolsof personnel managers in the private sector, but much of the early development'of examination techniques was done in the Federal government. The civil

service has also led the private sector in reducing employment discrimination

P rLF

0

-62-

4o

by race and sex (see Smith, 1977). Here statutory coercion has probably hadmore to do With any progress by American businesses than has the government'svisibility 4s a "model employer." The imitatilkometimes runs the otherdirection. For example, the Pay Comparability At (Public Law 91-656, 1970)pegs Federal salaries to compensation in the private sector; it is ond of manyCongressional attempts to pattern Federal personnel practices after those inprivate business..

More.abstractly, xivil service systems' approximate. the pure form ofbureaucratic. administration. "Bureaucracy...it fully developed...only inthe modern state, and in the private economy only in the mostragvanced institu-tions of capitalism" (Weber, 1968:956). The Federal civil service therefore ,

shares important structural similarities.with any other-large, bureaucraticallyorganized administrative apparatus (cf. Kanter, 1977). I6 particular, wherebureaucratic "rationality" underlies an organizatIon's personnel practices, theinstitutions governing the organizational labor market wili resemble those inthe Federal service. Fhe more bureaucratic an organization, the closer should.be the similarities. For a private enterprise; -civil service careers should

javd most in common with careers in the administrative component, which isusually more bureaucratically organized than the production component (Meyer,1980).

The results already reviewed lend support to these expectations. Our'findings with respect to human capital hypotheses,'our status attainmentanalysis, and our estimates of the amount and direction Of mobility acrossoccupational lines tend to be consistent, at least in broad outlines, withfindings for the general labor force (e.g., Blau and Duncan, 1967; Sommers andEcit', 1977). Like Warner, et al. (1963),.we infer that the careers of U.S.

servants are not greatly differdnt from those of their counterparts inany large priVate'business.

. Products and plans. We/boqclude this report by briefly noting someof our efforts at disseminating the results of our project, as well as the prq-spects for continuing, related work.

.

. wTo date, this project has generated major articles in the American ..,

Sociological Review (Tayl r, 1979),and the American Journal of Sociology(Grandjean, 1981),, plut paper in the California Sociologist- (by Taylor and . i

Kim). We -have presented apers at several professional meetings, includingthe 1980 American Sociological Association meeting, the 1980 Seminar onSocial Stratification of the International Sociological Association andthe Conference on Pay Equality (1079). More informally, we have discu.ssedour work with scholars in several" universities, and with employeeS of the Officeof Personnel Management. With supplemental funding from the U.S. Department ofLabor, we were able to undertakg additional analyses and supply that agencywith the report included here as Appendix B. Additionally some of our research hasbeen read into the records of Congressidnal Committee hearings on Equalityof Opportunity> .

..

..

., We view both the project and the dissemination of its results as on-,going. We anticipate several additional 'articles, and perhaps a monograph

presenting our results. We have received substantialawards of computerfunds from the University of Virginia to continue our analyses, and may seekadditional externalfunding as well. One M.A. thesis based on the projectdata is nearing completion, a second is in the planning stages, and. a Ph.D;s3

-63.

dissertation is well underway. We are optimistic that the present-"FinalReport" is in.fact but an interim report on a long -term program of research

. on careers in the U.S. civil service.

Mm

MOO

re

ors

. s

WS'

A-1

APPENDIX A:

HYPOTHESES AND OBJECTIVES

,:,}Duman Capital - formal and. on- the -job training:

}'pothesis 1. Years of scholing and amount of experience will bepositively associated with salary.

Hypothesis 2. The associations in H(1) will be stronger for white malesthan for non-Wite males and females.

Hypothetis 3. The highest returns to schooling will be for those employ-ees who have high levels of education and high levels of experience..

Hypothesis 4., Returns to schooling and to experience will be similar inmagnitude.

Hypothesis 5. Controlling for years of schooling and government experi-ence, earned degrees will be positively associated with salarlai

Hypothesis 6. Duration and expense of job training will be positivelyassociated with years of schooling.

II. Mobility - promotions and turnover.

Hypothesis 7. With entry level controlled, years of schooling will bepositively associated with,promotion.

Hypothesis 8. With education and grade controlled, there will be a

,, negative association between promotion and age after forty years of age.

Hypothesis 9. With grade controlled, years of schooling will be posi-tively associated with turnover.

Hypothesis 10. Net-ofgrade and-job series, there will be no significantdifference by sex in turnover.

Hypothesis 11. Length of potential prdmotion ladder will be negativelyassociated with turnover.

III. Occupational Career Ladders

Objective 1. ,Whatoare those entry occupations which leadto subsequentmobility?

Oblective,2. What clusters of occupations make up career patterns inthat channels or lines of occupational change appear to be established?

r\es

A -2

Objective 3. What occupations-provide the ability to "bridge" the majoroccupational groups?

Objective 4. What are"thokiNskills and abilities required to progressthrough career lines such as those derived. from Objectives 1, 2, and 3?

NOTE: One objective specified in the or.iginal proposal to NIE, dealing withreturns to education Oy college major, yies, hot pursued because of excessive'missing data on college major, and is therefore omitted from this list. Someof the remaining hypotheses and objectives have been slightly reworded, re-flecting a developing conceptualization of the issues as the research progressed,but the major thrust of each is unchanged.

4.

13.

ts

a.

0 I

a

B-T

APPENDIX B:'DEPARTMENT OF LABOR FINAL REPORT

V.

Baseline Research on the Operations of the Federal Government Labor Market*Department of Labor Grant #A20-11-79-32

Extension )o National Institute of EducationGrant #78-0005

4

. .

*Since 'grantees conducting research and development projects under: Governmentsponsorship are encouraged to'express their' own judgment freely, this report.does not necessarily represent the official opinion or policy of the NationalInglitute of Education or the Department of Labor,.nor any of their personnel.The grantee is solely responsible for the contents of this report. This reportmay be reproduced, in whole ,or in part,.for useAv the Federal governmetit.

4-

4

70

t,

T

O

Summary

As this nation's largest employer, the Federal civil service occupiesan important and visible place in the national labor marlset. The Federalcivil service'is visible due to i4s size and attempt to be regarded as a

model employer. It is important nbt only because of the programs which are-administered by Federa)'civil servants, but also because the personnelpractices of the Federal government may have indirect 'impact on private sectorworkers and 'direct impact on state and local government workers.

o Given the high interest in and importance of the Federal government asan employer; we prepared an analysis on a limited portion of Federal employ-ment practices. Specifically, we have attempted: (1) to assess the patternsof income inequality by minority/sex group among blue-collar workers; (2)to discover the effect of education and training on the salaries of Federalcivil servants; and (3) to 6ompare,the patterns of pay etructures of minority/sex groups over time.

Using a one percent sample of Federal civilian employee records, we findconsistent and considerable salary inequality by minority/sex group for bothwhite-collar and blue-co116r workers. Additionally, we find that education,.on-the-job experience, and job training have positive effects on the careers ofccivil servants, although these effects vary by minority/sex status and time inthe career.

a 0

"f .

.

TABLE Of CONTENTS

B-3

Page Number

TITLE PAGE B-1

SUMMARY B-Z

TABLE OF CONTENTS B-3

LIST-OF TABLES- B-4

INTRODUCTION B-5

ORGANIZATION OF THE REPORT B-5

RESEARCH DESIGN AND DATA SET B-6

BLUE-COLLAR WORKER B-7,

HYPOTH'ESIS 1

HYPOTHESIS 2

OBJECTIVE 1

- .RAINING

B-8

B-10

B-12

ilYPOTHESIS 3 B-14

HYPOTHESIS, 4 6 -17

4OBJECTIVE 2 B-20

EQUAL EMPLOYMENT OPPORTUNITYff

-

. HYPOTHESIS 5 B-22

B-26OBJECTIVE 3

A REVIEW-0 THE FEDERAL CIVIL SEHVIC4AS AN INTERNAL LABOR MARKET B-29

CONCLUSfONS . , ... Y . B,31

ti

"APPENDIX J. B-32 No(

z .

a

r-List of Tables

B-4

Page Number

Table 1: Regression Results Predicting Salary,1977, from Education, Experience, Age, andTraining for Blue- Collar Workers, Entry1974, and On-Board 1974. B-9

Table 2: Regression Results Predicting Salary,1977, from Education, Experience, Age, andTraining for Blue-Collar Workers, by Minority/Sex Group, B-11

Table 3: Pay Structure of White-Collar Workersand Blue-Collar Workers, for 1977 Salary. B-13

Table 4: Private Returns (in Dollars) to Insti-tutiOnal Job Training for White-Collar FederalCivil Servants, 1974-1977,.by Employee andMinority/Sex Group.

Table 5: Returns to Cost and Hours of General ,

,and Specific Training, Net Of IndependentEffects, Entrants, 1974.

B-16

__B-19

Table 6: Total Cost of General and SpecificTratkiag, 1975-1976, by Employment Status,'Whit*--rollar Workers.

C

Table 7: Net Economic Detriment by Minority)Sex.Status, for. Four Entry Cohorts, 1963-1977,

Predicting'Entry and Later Salaries.Y

, Table8: Pay Structure of Minprjty/Sex Groups,for Four Cohorts, 1963-1977, PredictingEntry Salary. ,

B-21

B-24.

.B-27 -

FINAL REPORT:

B -5

Baseline Research on the Operations of the Federal Government

I\

Department of Labor Grant #A20-11-79-32,Extension to National Institute of Education

Grant #7.8-0005.

Introduction

In the 196T,s and 1970's, we witnessed a gi.owing concern with the tenetsof neoclassicalkUnomics, especially as these basic assumptions referred tolabor supply and demand (Becker, 1964; Thurow, 1969; Blau and Duncan, 1967).Human capital theory in economics and status attainment research in sociologyboth sought to identify those characteristics of individuals which ultimatelylead to differentials in wage rates or the incomes of workers. Such character-istics include but are not limited to age, work experience, education:jobtraining, vocational training, and race and sex, although these latter twocharacteristics are not necessarily related to skills and abilities. Althoughthese traditions have provided basic information on worker traits and wages,there has been a noticeable lack of attention to irregularities in the labormarket which would cause wage returns to vary, by labor market sector, by race,and by sex.

The research undertaken during this project has attempted to estimate theeffects of certain labor supply characteristics Within one labor market - -theFederal civil service. By isolating workers within on'e' labor market, we areable to separate to some extent the confounding influences of labor market'seg-ment, race, sex, and labor demand (hours worked, type of appointment, etc.).For example, the finding that women and minorities have lower returns to invest-ments in human capital has been well-documented. However, we do not know whetherdifferences in returns to education and other labor supply characteristics byrace and sex are due to, for example: (1) location in a particularly well pay-M.§ versus poorly paying labor market; (2) years of experience in that labormarket; ,or (3) job training provided by an employer. These are onlya few ofthe possible influences on income that are generally left unexamined ih studiesof the returns of investment in human capital, and yet such influences surelyoperate to elevate or depress wages. Our study of the Federal service shouldprovide baseline information as to the possible extett of income inequalityby race and sex net of labor market effects.

Organization of the Report

A For clarity bf peesentation, this report is organized into five main sec-tions. First, we will discuss the research design and data set used in ouranalyses. Throughout this section we will indicate the significance of certainlabor market characteris,tics of the Federal government and those which aremeasured in our data set.

The next three sections deal with the specific hypotheses and objectivesenumerated in our proposal to the Department of Labor. These hypotheses and

B-69

objectives (listed in Appendix 1) fall into the, areas of Blue-Collar Workers,Training, and Equal Employment Opportunity.

Finally, we end with a summary statement of our work for the.Departmentof Labor, and the relationship of this work to that undertaken by 'two otherteams of investigators researching the Federal labor market, NorthwesternUniversity (J.J. Couturier, principal investigator) and Operations ResearchIncorporated (J.J. O'Leary, principal investigator).

Research Design and Data Set

In 1962 the Federal government moved to computerize its payroll records,and began the Federal Personnel Statistics Program (FPSP), a lobgitudinal work-history file-on a ten percent sample of Federal civilian employees. The aimwas to provide a statistical basis for work force analyses which would cgntri-bute to rational personnel management. Over 'the years, 'changes in the auto-mated data system have improved the accuracy of-the records while increasingboth the amount of information stored on each employee and'the numbec,ofemployees in the file. In 1972 FPSP was superseded by the Central PersonnelData File (CPDF). Like its predecessor, CPDF covers virtually the entireExecutive Branch, except for White House staff, intelligence personnel, employees'of the Tennessee Valley Authority, and politically appointed heads of agencies.Included also are'the General Accounting Office, the Government Printing Office,the U.S. Tax Court, and the administrative offices of the Federal.cOurt system(see Schneider, 1974). Transaction histories are now maintained on 100 percentof these employees.

Ot

Monthly agency reports, submitted to

.constant,

Office of Personnel Management

updating of personnel actions occurlire:hroughout the Federal civil service.

by each Federal agency, have been with the CPDF to provide onstant,i

In the report that follows; we draw from that compilation of data. The bulkof the original data'analysis in the chairs that follow is based on theofficial personnel records of a one percent sample of Federal employees, drawn'primarily from FPSP and CPDF. Additional files, such'at the.Minority GroupDesignator File, the Retired Military File, and the Training File, were alsotapped for information to build c16r- longitudinal file referred to here as theFederal Career File, or FCF. This composite data filgwon Over 69,000 indivi-duals employed by the Federal government for any length of time between January,1963 and June, 1977 was made available for our use by the Office of PersonnelManagement. . .

The information included on FCF is of considerable variety. For example,on each Federal employee a status record as of June, 1977 (or as of theirseparation from Federal employment) inclddes informatiop on birth dates educe- .

tion, agency, geographic location, sex, minority group status, pay group, salary,veteran's status, length of Federal experience, and so forth: The dynamicanalysis in our study is possible from the information contained on transactionrecords for each employee. The number of personnel transaction records variesby employee, and in, principle has no upper bound. The minimum is one, forthose who had just entered Federal service in June of 1977 or shortly before,or whose first transaction in FPSP recorded their departure from the Federalservice. The'observd max'imum in-our sample is 64 for a long-term employeewith, quite obviously, frequdnt transactions. The average, is 8. Personnel

B-7

actions recorded insude, but are notlimited to, grade increase, occupation.change, merit pay, increase, salary adjustment, ,demotion, geographic change,agency transfer, furlough, military leave, and separation, with codes availableto distinguishthe reasons for a transfer, separation, or furldugh. Personnel

,actions can also include a change in.status, such as from career-conditionalappointment to career, or from excepted service to competitive. Many of theseterms will be discussed in later sections. Readers who wish more 'detailed

4F information should consult the Federal Personnel Manual, Processing PersonnelActions, and Personnel DataStandards, all published by.the Office of PersonnelManagement (formerly the U.S. Civil Service Commission).

4The final portion of FCF consistsbf information on job,training by

employee from 1974 through 1977,, a,ed includes cost and hours of training,purpose of training, type of training, and other related data. We are able,by the mariner in which these various data elements were merged into FCF, to

:link demographic characteristics of an individual with a history of personneltransactions and training instances.

For our research Eder this grant, we used the data from the FCF to assessthe effect of various individual worker characteristics (such as education,years of experience, etc.) on salary usi;g least squares regression analysis.Specific characteristics, or variables, of interest were operationalized andtheir effects on salary or trainingawere assessed net of the effects of otherconfounding characteristics such as age, sex, and race. While there are cer-tain limitations in this method of analysis (the specifics of-which are dis-c1Jssed later),. his method of analysis follows in the tradition' of human capitaland status attainment research, as well as being a suitable method to maximizethe detailed information/in our data set.

I. Blue-Collar Workersr

,-,.

Many studies'of returns to on-the-job training and education were fashionedaround the experience of blue-collar: workers, where increments to time on-the- ,

job and skill level were readily transformed'in hourly wages. It is therefore>fitting that we begin our discussion of the Federal labor market with the andlysis.

,

of blue - collar workers:. .

.

*,

AO 4.

*In. the Federal civil service,lolue-collar workers comprise about 30'percentof allcemployeet and are distributed ab-oss the same geographic areas as- white -collar workers. While many of the blue-collar workers,arq0ssaciated withitfieoperations of the military and are found in such occupatlirs as aircraft main-tenance, aircraft supply, et:., the Federal government also has manyApositionsfilled in occupations stch as supply clerk, dispatcher,, and food serlick Infact, there are more bluelicollar job titles than there aPe white - collar titlesin the civil service. ,,

t)' . . ,

. .,.,'

, , ,

. - / , -,:

Similar to the agency and geographic dispersion of white-collar workers',..

bluelcollar viprkers are goierned by many of the same personvl regulations thetgovern whip-tollar workers. Although the pay systems of Are-collar workers ,

are more varied, than those otheir White-collar counterparts,' they,are nonethe-,

less:subject,to similar requirements regarding suitability for employment,regyisitelevels of:edwcation and experience for the job being filled, rates of .,

advancement, and so folly'. Therefore1 status attainment and human 'capital models`,

. . .

.<°0.- ; t.

B-8

developed to fit the white-collar employees night likewise be representativeof th'e earnings function of blue-collar workers.. ,

As an amendment to our study under_the National Institute of Education ,

we develOped three hypotheses and one objective (question) regarding blue-coillarworkers. First, we,wished to establish that similarity between the Federal'government labor market and the private sector with regard to returns tq school-ing and experience among the blue-collar workforce. .We therefore positedthehypothesis that: , ,-

-'. ,

H(1) .Years of schooling and amount of'on-the job trainingwill be positively associated with salary among blue-collar workers.

. -

Using regresgiOn leastisquarps procedures, we 'regtessed satary'in 1977 ontotwelve independent variables for a cross section of the blue-collar Federal .,

work force. The.results from this, analysis are,presented ijn'Table 1. - ---

,,,Pre.se(ted-in Table 1 are the regression coefficients for each Of the in-

,. 1/4 ...

, . : ,. 4 .

depenOnt variable's preditting 1977,salary,for two groups of blue-collar workers.We examined both emOloyees,who entered in 1974 and were still on -board in 1977,and those employees who'were on-board in 1974 and still employed inthe Federal ,

goveriPPent in 197:7,egardless of when they entered. Tnetregressiun equations-predioting 1977 salary use the same variables for both employee groups with theexception of years'af,Federal service.- In the 1974 Entry cohort, years .ofexperience wasinot varied since these employees all entered at approximately, the,same time.' Thtrefore, YEARS and YEARSSq.. were nOt'entered into the analysisfor the Entry group.

..1

.-' .

.

...",As to be expected,' ducation has a positive effett'qn.salery,net of the

skill Categories, age, and minority/sex group. For.eachyear of schooling,blue,-collar workers who entered in 1974 received $216 on the average. Curiously,

.

-however,,:hours,of classroom-style job training which they ,received (TRAINING):.

is not. poisitively associated with salary,. For` every hour of-job training re- .

ceive Pie entry blue-collar worker received no-noticeable.incrementto theirsalar . '

,-, .

.

i...

When tie results of the,on-board.group are studied, we note that 411, typesof skill-producing ctiaracteristics are positively ?elated.to salarw. for teach

year of education aremployee has; there was a return.pf $315 in salary.Similarly, .yeats of Federal ,employment experience was, ,positfvely related. to1977 salary (b =.1189) as was age of the employee (b = $211). Significantly .

different from the entry cohort; however, is the fact.thlt on the-aqrage, the' .

on-board employegs received $2.33 for each hour of classroom style trai,ningwhich they received. This finding suggests'that the training, received by,,blue-' - ;

, collar .Workers earl in their tenure'Witfi the rederaltoverliment is.nortratpingr .

which can'.advance.their carders, but rather is training necessary 'for' them to: ' _.:

.so-work at Me.minimal level in their job. -For. more senior wotters, however,.such training canlbe used tb augment their earning capability; 140allowinglIhem

-..

tO"Move ahead`-in their*.job, to move into supervispr pos.ttion5, or to make 'occupation or agency shifts. , t . .te

4 .t , .v*------. .

7

'Consistent with findings on white-gollar ww2kers,--minorities and womenearn Considerably less than white males. for both, the.,entry.gr,oUp and the on-,boardgrOOp (see Taylor, 1979; znd-faylor and Grandjedn,'-1.979). -' - '-'

-.

, 1-.

e -

B -9

Table 1: Regression Results Predicting Salary,*1977, from Education, Experience; Age,and rraining for Blue- Collar Workers,Entry 1974, and On-Board, 1974.'

4160.R

. o 4, )Unstandadiik Regression'Iridephdent Variabl,es2 -,

, Coeffftients -.

.4-- 4 t,Entry, 1974 ,,

Etcatioai 216*

ekrs-

, Years q. - .-

Age , 1 .44

. Age'Sq. .or -1,79

Training - ...D 0'N .....

Minvity Male' , % -425

NorilMinority Female - ' --i595*

Minority Female -17.35 **

R2 (adjusted): .34;

N ''..-Pt4 53

0

\.

On--.Board, 1974

315 * **

789***'--

-2.39***

211***.

.-a.25***

2:33***

1898 * **

71440***

.341 1r

.2824

'For these still on-board as of June, 1977..

2Net 0f,dummy.variablesgkor preMium skilled, skilled,.semi-skilled; andunskilled. jhroUlghout this table and the 'rpmiinjng text:minority employee

refers to these employees Who were designiled as Negro black, HispaniC,Ameran Indian,or Asian/Oriental. '4,

A-

r.

I

*

1

.'7o

o

4 .

o 4

O

.'

A

B-10

in sum, these results suggest that blue-collar workers in the Federalgovernment show earrKings functionS similar to blue-collar workers in theprivate sector, et least in broad outlines.

.

Giwen the significant pay differences between the minority/sex groups asi

a.

llustrated in Table 1; a logical question which flows from that finding is,are there djfferences in the pay structure for the minority/sex groups whichaccounI'for.differences ia.salary levels? This brings us to Hypothesis (2),

°

H(2) The assocfa on in H(1),will be-stronger for non-minoritymales than for any o her minority/sex group.

We reasoned that salary differences among groups of workers are due to theperceicied quality of schooling,-experience, etc., and that such quality wasperceived to vary by minority/sex group. Therefore, even if minority males,'for example, had as high a mean education level as white males,linoritieswould not reveive as much in total salary based4n their educatilon as whitemales.

In Table 2, we present the results of regr sion analyses. In this table,an incomedetermination function was fitted to each minority/sex group separ-ately. Therefore, we-allow for the interattiv fects 'of race, sex, andemployment characteristics.

The results show greater similarity forthetwogroups of males than we had'initially expected to Mind. Forexample i,training is significantly related tosalary for non-minority and minority males only. Similarly, educatiori issignificantly related to salary for both groups of males rather than for eithergroup of females. An although years of Federal work experience is positivelyrelated to salary across all four groups; the 'returns to experience are moresimilar for non-minority and minority males as compared to any other two, groupsof employees.

The coefficfents for thetwo groups of females while not significantly.

different from zero, are generally in the same direction as that of males.This is not true, however, for age for non-minority females and training forminority females, coefficients which doliot attain statistical significance.

We should note,1hoviever, that these tentative conclusions on the'paystructureof 'employees in the Federal blue-collar-sector are based on rela-tively small sample sizes for both groups of women. Significance levels maybe more difficultto sustain with the small numbers of females. At any rate,the initial hypothesis which we put forward does not appear to be substantiatedby these findings. 1

On the other hand, it could be argued tht the skill group control mayhave depressed the returns tc human capital investments. That is, white malesreceive greater returns to education, etc., because they are able to placethemselves into higher paying skill categories. In lower-skilled ilork cate-

gories, the pay structure might not be flexible enough to allow for variabil-ity in educational attainment. Hence, the returns to education would bedep.resied, whereas seniority, for example, might have a greater impact'on

. salary. We performed such an analysis (not presented here) and concluded

70

,

A

1 I

Table 2: Regression Results Predicting Salarjr:1977, from Education, Experience; Age,and Training, for Blue-Collar Workers,by.Minority/Sex Group.1

oor.low

Independent Variables2

Unstandardized Regression Coefficients

Non-MinorityMales

MinorityMales

Non-MinorityFemales

MinorityFemales

Education 341*** 273*** 351 115

Years 218*** 136** 285* 481***,

Years.Sq. -3.12*** -1.80 -6.51 -12.71***'

Age 253*** 108 -67 17

Age,Sq. ,-3.14*** -1.47 1 -.72

Training 2.32** 3.39* , 19.35 -1.0

(cons-tant) 114050*** '6205*** 6728 8538**

' R2 (adjusted) .266 .233 .289.

N 1826, 778 85 135 '

1For the on-board sample as of June, 1977.

-Net of dummy variables for premium skilled, skilled, semi-skilled, andunskilled workers.

J

80

B -12

again, that non-minority and minority males are still more similar to eachother than'to any other group of employees. However, the returns to thehuman capital variables.of education and experience did attain statisticalsignificance for,all four groups, suggesting that skill level platementmarkedly affects increments to salary which woyld be Venerated by educationand experience, especially for women.

Interest in pay structures has in part been generated by the continuing -

differences in salary between rate and sex groups. Part of our interest inthe analysis of the Federal -government Is the comparison of b1N-collar a"

white` - collar workers. Given.some similarity in the types of regulations ichaffect all Federal, workers, as well as similarities in the manner in whic theregulations are appli ed, we, would expect that there would be siMilaritiesin.the pay structures of blue- collar and white-collar workers. As one of ourobjectives in this research, we posed the queslion:

0(1) Doe the pa;*structure (i.e.., the determinants of earnings)of blue-collar workers differ.from that of white- collar workers?

We performed separate regression analyses on a sample each of blue-collarand white-collar workers, using the variables outlined earlier. We alsoentered dummy variables for'the minority/sex groups, as well as duinmy variablecontrols for the occupational and skill groups {see footnote 2, Table 3), Foreach sample, we are comparing full-time employees.. who'were on-board in 1974

remained on-board until June, 1977. Salary in 1977 is the dependent'variable.

There are five results from the regression,analysis which appear espec-ially noteworthy. First, the variables used to predict salary ,explain more ofthe variance for white-collar workers than for blue-collar workers. This isin 'part due to the explanatory power of the-occupational group categories Usedas control variables for white - collar workers, as compared to the skill tate-gories"used in the blue-collar'regression.

Secondly, the returns to most human capital variables are larger forWhite-collar workers than for blue-collar workers. For example, each year ofschooling receives on the average $565 in alary increment among white-collarworkers,, but only $315 for blue-collar wor 'rs. These differences, however,are in part due to differences in''Oe.sal y ranges for each group .of workers.When the elasticities were examined; we found that the average percentreturns are much more similar between the two groupS1Of employees (data'notpresented here).. .

A third result of note, and one perhaps most critical,to assessing the

degree of similarity between the pay structures of the two groups, is that theudering by size.of the human capital coefficients is identical between the twogroups. Education appears'to have the greatest salary returns, while classroomstyle training has the least. This finding suggests that the pay structuresof blue-collar and white-collar workers are generally similar, at lea9t.accord-inOko the variables measured here.

.110K

Of note also is that classroom style training does not appear to haVe a4

significant impact on salary for white-collars once occupAional group iscontrolled. That is, occupational group, net of education and race/sex group

81.

B-13

Table 3. Pay Structure, of White-Collar Workersand Blue- Collar Workers, for 1977Salary.l

Worker Group, .

.:-. Independent,

Variabges2.

White-Collar Woiske.rs Blue-Collar Workers

Education 565*** 315***'

Age. .. . 225*** . 189***

Age Sq. -2.56*** -2.39***

Years 281*** .(/- 211***

'Years Sq. -3.41*** -3.25***

Training .... .04 . 2.33***. . .

Minority Male ; 999 * ** . -\952***

Non-Minority Femdle, -2938**,* . .-1898***

Minority Female -2656t** -1440***

A2' (adjusted) . .633 .341

N l0,460. 2,824.

/

1For all workers who were on-board in 1974 and remained to.1977..:

2Net of occupational group. For white-collar workers, these.groups are pro-,fessional, administrative, technical, cleribal, and other. For blue-collarworkers, thesegroups are premium skilled, skilled,"semi-skilled,'And un-skilled. -r

.4

. .

.

4

4

fr 1

'B-14

for example, determines both the amount of training one receives as well asf7'

determining salary for white-collar workers. Fortblue-collar workers, .however,training has a substantial impact an 1977 salary.

It

Finally, salary inequality by race and sex is apparent for both groups ofemployees. While minority males earn significantly less than non-mind,itiy -

males, both groups of female employees earn significantly less than do minor-.ity males. And, in fact, the minority/sex groups fall in the same order ofsalary inequality vis-a-vis non-minority males for both white-collar and -blue-collar employees.

Given that the employment experience is different for blue collar andwhite-collar workcers, it would be difficult to assess all employment cha'r-acteristics.affecting salary. .The.results here suggest, however, that the paystructures are similar-in kind, if not im degree of returns for both groups ofworkers.

The difference between the'two grbups of employees in the effects ofclassroom style training is curious giVen that there is considerably moremoney and time spent on training among the white-collar work, force than among

,

the blue-collar group.

We now turn to an examination of the returns to classroom style trainingin the .Federal government. 0

II. Training

Classroom style training differs in a number of respects from on- the -job'training, or OJT (see Taylor, 1980). Job training which has been institu-tionalized into a classroom setting, away firom.the work site, we refer to hereas Institutionalized Job Training, or IJT.

\ .

A major concern in this research is the relationship bf returns totraining by minority/sex group. As we saw above, for a Cross- section ofwhite-collar employees there is no significant salary return to training. In

'part, that finding could be confounded by: (1) the use of hours of\trainingrather than cost of training; (2) differences in returns to training acrossemployee cohorts with younger employees experiencing Treater returns to

training but old effiployees receiving little benefit; and (3) a different.rejatippship of training to career advancement each minorfty/sex group.

These concerns led us to an analysis of returns to training by minority/sex group, and we-hypothesized that:

H(3) Returns to training will be greate'r foraion-minority malesthan for any other minority/sex group.

-.

To estimate whether there are, in fact, different monetary returns to JJT, anearnings functiOn for each minority/sex group was estimated using 1977 salaryas the depen ent variable. Cost of IJT was entered as a predictor variablein regression equations along with additional control variables specified inModel 1 (educ tion, age, squAre of age, and for the on-board population,seMority.and square of seniority). Model 2 for each minority/sex group

83

B-15.

introduces the control for occupational stream, while Model 3 adds pre-,training salary. The results from these analyses are presented in Table 4.

Among the 1974 entrants to Federal employment, IJT provided the greatestreturn% to non-minority males under Model 1. However, with additional controls

.

for PATCO and pre-training salary, the returns to the dollar investment inIJ-rdrop,dramatically for non-minority males, but the controls for PATCO andpre-training salary actually increase the returns for the other three minority'sex groups. According to the results from Model 3, the 1977'salary of non-minority males was increased an average of 48 cents for each dollar theFederal government invested in their IJT. However, minority males received aprivate return of $1.01 on each dollar invested in their IJT; non-minorityfemales, received a return of $.98 on the average for each dollar invested intheir IJT, and minority females received a return 00.90 (although statisti-cally non-significant). The higher rates of return,10 IJT for minorities andwomen among the Entrants suggest that their salary -levels could be improvedvis-a-vis non-minority males if they were given equal ,access to IJT.. Forexample, white females had a rate of return on IJT of $1.00 net of occupa-tional stream, pre-trainino salary, etc. If they had been given as much ,

training as non-minority males ($269) rather than their own $9O worth oftraining, their average salary would have risen by $179 more than jtdid. Sincenon-minority females' salary increased by-$1,607 during the period covered here, -

the 5179 in additional salary generatedby increased access to training wouldconstitute an additional 11.1% increase in salary.

\''

.These findings are somewhat surprising given'.the-data presented byMincer (1962)-and others where the estimated returns to OJT viere greater forwhite males than`qor other minority/sex groups. However, introduced in thisanalysis is a'ebntrol for occupational placement and employer,'as well as amore precisely defined cohort of employees who entered employment over a one-,year period. We shouldaadd that this same pattern of results is obtSinedwhen number of bolirs.of training is used instead of cost of training. Thesequalification necessarily make these analyses different from earlier esti-

\.pates of the effect f training, along with the different type of trainingbeing consid red hey e (although Rosenfeld, 1980, obtains similar results).Moreover, the e -aria yses should not be taken-as definitive for employees ofother labor mar, s, where rules governing competition, promotion, etc, maybe very different. As a "structured labdr riorkeel(Phelps, 1955), thre-sponse of the civil service to the supply o1,workers may be diffe'rent from'.More open and unprotected markets. However,'this should affect only the

,. magnitude of the returns to training rather thaR the pattern of the - 4

tionships between minority/sex groups.

fewer job. s is fort the trainees to_move. upward., and into. This is consistentwith studies of OJT that report'a declining rate of,return. ,as the employee

. f 4, .'When the on:bprd population/of Federal employees is analyzed, we find

a drop in the retiirns to IJT for all'minority/sex grota of employees. Infact, in Model 3,kthree of the foUr.coefficients,are negative,afid the fourth(minority males) is non,significabt. in part, these find,ings can be explainedby three factors. First, these employees have been in the Federal service foran average of 11.5 years and have therefore been moving along their carerppths for a considerable time, .This would that there may 'be a limit tofuture job moves since they have already clos dsome career options.., 'That is,. - .

there is le s of a chance 'ford return to training simply because there are'

1\

4

B-16

Table 4: Private Returns (in Dollars) to Institutional,JobTrainingfor White-Collar Federal Civil Servants,197471977, by Employee and Minority/Sex Group.;

Minority/Sex Group

Model laN2p-Minority 'tales 1:77*** .65***Mnorfty Males .20 ' .90***Non-Minority Females .g8* .94**vMinority Females .04 .48#

, -

Employee Grail

Entrants, 1974d On- Board, 1974d

Model 2bNon-Minority Males !Nimiliti4 ...24***''

Oinortty Males - .16 .41**Non-Minority'FemaltSMinority Females ,

Vogel 3cNon-Minority MalesMinority MalesNon7Minority Females

. Minority FeMales

.72* .21 , .'

1.23 :04

.48*

1.01# .07

.98*** -.50***-

;90 -.50***

a: Variables entered into the mOdel.are education, age, squareof age,length of Federal service, and square of length.of service.

.

. ,

, ,

1 ' : Variables entered in the.model are.Oducation, age, square-of age, PATCO,1 the minority/sex variables, length of Federal service, and square of .

length of service. .

.... .

. .,

c: Naridbles entered in the model are education,..age, square of age; PATCO,pre-training sajary, the minority /sex variables, length.of Federalservicp and square of length of service.

d: -Levels-of statistical significance are: 4, p ', :l; *, p < .05; **, p.< .D1;

-

.and ***, p < .001.. .

. .

O

1

B-17

ages. Second, however, is the fact that sopne training in the Federal serviceis given only after an employee has advance-d-to alcertain level, and the higherthe level, the more likely this is thecase. For example, the three trainingcenters for middle level managers and staff personnel are open only to, thosepersons who are. in General Schedule grades 13-15. The Federal ExecutiveInstitute is open only to GS 16-18's (and a few selected GS-15's): Moreover,all GS616's must-attend the FEI within one year of their Momation into thesupergrades. Thetefore, some of the most expensive and extensive IJT comesafter .a promotion for the longer-term employees. It might be argued that IJTfor these employees. results only in enterprise returns rather than individualreturns, That is, the increased productivity of an employee brought about byimproved skills from IJT are "captured" by the employer, rather than returnedto the employee. Finally,- it could be that formal job training mid-waV4cough a Federal' career may be given disproportionately as remedial OJT; that

s; IJT is given to older employees who have somehow missed the lessons frpmexperience on-the-job.

'While job training is certainly distinct from education, and IJT as wehave shown here is distinct from OJT, at least two types of )ob training havebeen distinguished by Becker (1964) and ofhers in their ana hses of labormarket experience. General job training is regarded as-skills developmentwhich is transferable from one employer to another. Courses or instruction inreading skills, general factory procedures, or equipment maintenance would beexamples of generaltraining. ,According tee the human capital model, employerswould be reluctant to invest in general training for their.employees sincesuch skills are employable elsewhere. 'Should ah employee be trained in skillswhich are transferable, then the employers might,easily lose their investmentin the employee from attrition, or be forced to pay more in wages 4io keep'

.generally trained employees from leavings, On the oter hand, specific trainingfor employees shot& be sincepreferred by employers sincspecifit training, by

not-vtran 43' raPle from one employer to another. That is, trainingin a computer system ope#, ed by only one company, or training ,in an occupation,which.is specific to one employer, ,or training in a manufacturing process usedby only one firm, is said" to be specifi.c.tn na4ure. St.rice'the.employee is °un111.(ely tope able to use such skills in any ogler employment context, theemployer'ngednot increase wages.based only on treini g. -

ThiS, of course, is a rathei°abbreviated treatmen of pecker's work ongeneral vs. specific training, and he himself ackn9w1,edges that the,dis-

.- tinction between general and specific training islhat of opposing ideal

,1

types (see also Lloyd and Meimi,.1979:123): kotetheless, such'a theoreticallyappealing divis.ion'af job= training suggests, mud about the response of employ-

ers and to training, and makes assumions `bout the relative worthof trained versus untrained-employees. Therefore, we hypothesized that:

jr H(4) Individual salary returns 'to geplialbtyalting will exceed-1/4

those to,specific training . I / 44,. , .4 .

Heretofore, the distinctijn between general.and specific training'hasbeen based rgely on theoretical) mathematidql models, since therelhas beenrelatively the inveftigatidn of the effects of various training pcograms'by type if training (i.e.,-general or specific).0But, there are somd s.ub-stantive reasons to suspect that a distinction be teen general and specifictraining"mighenot have economic validity for al abor: Markets. First,

-,

4, ilk**

88

1

B-18

employers not only invest cost of training in their employee's, but they investtime. as well--time separ;te from that which is related to the cost of training.That is, employers must manage their labor Pools to keep suitable, qualifiedemployees in line for openings which occur in their-firm or agency.

Since recruitment from outside the fine is likely to be costly and time-consuming, employers have an incentive to increase the wages.of trainedemployees to keep them with the firm, especially if the training is specificin nature. That is, direct training costs are only a part of the total train-ing, cost. Secondly, and as Becker notes, some "specific" training includes ageneral component; so that skills learned in, and for, one setting can e

elaborated on or refined to fjt4onother employer. Training may be morespecific than general, and yiceversa, but it would be difficult to separatethem completely. Thus, any on-the-job training should be useful in increasing,productivity and in increasing wages. Thirdly, the human capital model seemsto ignore a credential effect whch.could be geherated by training courses,whether they be specific or general. The persons who receive training are notonly assumed to have more skills after training but may also receive soreincrement to salary simply because they were deemed deserving of training ex-penditure. Also, most persofts have an expectation, that training incritases theirworth to an employer, regardless'of whether the trainina is spedific or general.Given an expectation that their wages should increase with training, there willbe pressure to'increase wages after an employee has received any t-aining.

By its method of record keeping, the Federal government makes possible a

distinction between general and specific job training, Using the same trainingfile as de, cribed earlier, information on 'pyl-pose of training is kept f:, eachinstance of training. General job training; that is training not specificothe Federal government, included theyfolfowing: training cours or ne4technology, to.iTprove Present performance', to meet future stafTingneeds.,,astrade or craft apprenticeship, and for adult basic education. Specific.tretn-ing'inctudes: training for a new work assignment, for a program change, todevelop unavaillple skills, and for orientation to the Federal government.

'These two types of training are operationalized as traeig which is,eitherspecific to theFederal .government or not, rather thanas training specific toan occupation: The definition for these types of 'training are derived fromBecker (1954). The particular ones used bk.the.Federal 'government are reported-in the notes to Table 5. .

,

For both specific and generVtrainfng,.regression analyses were under-taken where 1577 salary was regressed sepkrately onto cost of either 5pAcificor general training received.during 1575 and :1976, and hours Ofleither specificor.gencral training. The results from'these four regressioneq.uatibns.for the1974/ Entrants are reported in Table 5.' The effects of training'on-salary arenet of pre-training salary, and eight additional independent variables. Thefindings are somewhat surprising. While human der a1 theory predicted thatreturns to general training would be greater than returns to,Specific,tra*-

. ing, the analyses here present mixed results. Cost\ of.specific training hasa greater individual return than.doeS cost ofgener0 -tra.ihtng, although thedifference between the two estimates is statistically nonsignifiCant. The .

reverse pattern is found for hours of general and hours of specific training,r and similarly the difference of the two coefficients is statistically no

significant. In all cases, however training isyleasuttd, ,it 'is positiverelated to salary, alhQughthe leve of.significance is'only p < .10 fohoursof specific training:'.

.

a8 7 \

. 1

)0.

8-19'.

.

Tab)e 5: RetUrns .to Cost and Hours of Generaf'and

Specific Training, Net of Independent-Effects, Entrants, 1974a

Measure of Training

Cost

Hours

Generalb

Type of Training

-Specific

.66*** .80* .

4:47*** 2.22#

aNet of eduction, age, age squared, Fede\-al .service, Federal Service

4'

squared, PATC0i minority/se% group, agency, and pre-training salary. Levels.of of the coefficients are: *" = p<.00T; * = p <.05; # = V.10:

bGeneral training is defined by' the following purposes of training:

new technology - -to Provide the knowledges or skills required.to keep abreast of. developments in the employee's occupational field, or in a related

field. (State-of-the-Art)improve present Performanceto provide the knowledges or skills needed to

improve or maintain proficiency in present job:' future staffing needs--to prOVide the knowledgesvd.skills -needed to meet

future staffing needs through a planned career, development program. 'ttip aw-occupation' specialty': ,1 . _

t

ctrade or craft apprenticeshipto provide thesclassroom.0 group portion of

the-formal training-that , together, with guided work experience,.Armit the employee to acquire the knowledges and skills needed tpmeet the full requirements for journeyman status in an apprentice-,ship program.

. . ,..

adult basicleducationto provide the basic knowledges and skills needed tolir permit the employee to.function in the world of work.

..

7-.

.4cSpecifIc training is-defined by the followtna purposes of training:newrork aSsignmentto provide the knowledges and skills needed as a result

of assignment tookw duties and, responSibilities, when such trainingis not a part ofia)pTanned career development program.

. prograth change - -to providethe knowledges:or skills. needed as a result ofchange in agency; ission, policies, programs, or procedures.

develop unavailable skills--to provide the mowledges or skills needed for. fields of work tRat are uhique to the Federal Government, or to

meet Governmentstaffing needs in oecuPations for which the labormarket cannot produce a sufficient number of trained candidates.

orientation- -to provide orientation to tfie,Olicies; purposes, mission andfuhctions Of t'i'e employing agency or the Federal Government for newemployees. , \ v ,

."..

1.

\ 8 S

4kAIL

INF

01.0/11,11,

4,

B-20

These findings suggest two important coniclusions. First, specific andgener'al training are important for new entrants as we saw earlier. Butsecond, speciflc training obtains monetary returns at...least equal to generaltraining. Thus, scmewha't contrary to the conclusion drawn by Becker andothers, an employer (i.e., the Federal government) does remunerate employeesfor specific training even though such training might not be used by anotheremployer. Concerns with this part of human capital theory were expressedearlier and need not be repeated.

However, a question naturally arises as to wh..the Federal government,might vary from the expected pattern of returns to soeCific and generaltraining. First, it should be noted that the Federal government is.a verylarge employer. An employee could move from one agency to another averreceiving training. Hence, an employee could obtain significant returns tospecific training by changing agencies for a promoti-6 after receiving specifictraining in the first agency. To the extent -that this :argumerit is valid, theFederal government faces the dilemma of a bureau'or division paying for employeetraining only to have theMployee hired away by another bureau or agency.There are sore provisions against such moves in tra4ning agreements, however.Usually an employee who receives I,3T sigps a statement that he or she will notleave the first agency for a Specified period of time, usually three tosixmonths. This policy suggests an attempt to inhibit agency shifts by recenttrainees. However; this limitation would probably not be very eff-xtive inlimiting job changes across agency lines for a very good reason. Proceduresfor Job changes which occur across agencies are likely to be initiated onlyafter a trainee has.made contacts during a training session.. Given thelength of time required for job vacancy announcements, panel review of applica-tiohs; interviews, etc., the trainee has more than likely repaid anytime'owed to the first agency,

All in all, the findings presented here necessitate a revision of therelative returns to specific versus general training by the size and organi-zation of an employer. That is, large fir'ms may show significant returns tospecific training.whereas small firms do not since employees, can move among

,major*firm subdivisions.t

Another'related issue in t he study of job tra ining is the notion i4tsince general training can be Used by many firms, those persons who receiveg,eneral.training may be morelikely to leave.a firm than those who receiVespecific training. Given the concern for fiscilreSponsibility and costaccounting in the Federal bureaucracy, there is a more generalized concern thatpersons who'are likely to terminate their employment do not receive consider-able amounts of training funds which' would then be lost to the government asan employer. 'Concern with such issues brings us,,to Objective 2:-

0(2) What is the annual cost of general specific .

training in dollar outlayor employees whO,remain in theFederal service and for those who

Using the FCF, we estimated the cost of training by type of training fortwo groups of employees: those employees Who entered Federal service in 1974'and left by 1977, and those employees who entered in 1974 but were still on- ,

board in, 1977. Presented in Table 6"are,the sample estimates for rscost ofclaSsrgora sty-le training 1975-1976.

8J, b

o

I

B -21

Table 6: Total Cost of. General and SpecificTraining, 1975-1-976 by EmploymentStatus, white - Collar Workers.'

Type of Training2

General'

Specific

Cost of Training by

'Entered, 1974, andTerminated by 1977

(N = 9377-.7

S2,801

S 577

Employment Status-

Entered, 1974, akRemained to 1977 f

(N = 691)

597,217

$18,746

1 1

-The termination group consists of all persons regardless of hours worked,career status, or service. Those who remained with the Federal gcrernment are'full-time employees.

2See footnotes b and c to Table 5 for explaation of types of training.

d

r I

. .

e

b*,

,... :-; ;'.

^ i

I) 1.P

y,,

, , -.....

,,, ,. v f- -., . .

.T . % *The figures are only rou,ghestimates as to-the oast of training Sincein'tra-agency training is not 'in'cl'uded, nor is employees' salaries figured into

.

the cost. Hpwever, from the data in Table 6, it appears that Terminators erefar less likely'to receive Federal funds for IJT. Per perGon teWhato-rs Areceived less money than the on-board employees, as well as receiving fewer

-, ,total funds as ,a group. And, tn keeping with the tenetS'ipf human capital 'a .,.,, theory terminators receivedjeWerfunds for specifid, training than for 4 ..11,e.4

)4eneralz

training... Homeverf'the same.can be said of the on-hoard sample, 4here' '

we might have expected more funds spent on specific trainiff§\than on over1training: . . . .,, ..

:*0. .. -7 04,'

r 4. , 4

,

4 ...

% 4 ,Clearly, more investigation should be done in' thiS'aree. 'Our particula,

-Sample is not as'well suited to answer Objective 2 as is desired. Questions.need to,be answered ,regarding the types of employeesOlo-yeceiVetrafncing and, ...

t leave. ,Why did they 'leave Federal employment; Are the.terminatoft obtaining . :!..'obs.in the private sector? If so, what is their' salary? The records, the. ,,,

Officeikf Personnel Managerlent is collecting,cannot answer :' these and atilt' ..-

impqgtit questions regarding resource litilization,and eimplOyment status. The! '.answers tosucn questions may better aid managers in the allbcation of qa4ninft A.,,-

, .,funs fox more effective personnel planning,, .

q..

-,.

III. Equal ,Employmel t Opportunity,

. .

dur third area of investigation under"this grant was the astessment.ofincome inequality over time by race and sex group in the Federal government.Earlier research by Taylor; (1979) pdinted to large income gaps betweenminority /sex, groups even when differences in emplbymeni profiles such as edu-cation, age, and lxperience were statistically pntqllesi. However, given theefforts expended on affirmative action Programs,,the negative connotation'associated with employmentAiscrimination, and azveneralizedbelief-that.women and minorities have'been:discriminated in the past, we should find a.%decrease in earnings gaps between the minority/sex groups. We thereforehypothesized that: ,

I ,

H'(5) Among white-collar workers, salary discrimination by raceand sex has declined from 196Z to'1977.

Using the brie ptrcent sample of Federal civilian employees divided into

four entry cohorts, we perforMed regression analyses' using salary at threeseparate time9 as the dependent variables. Thes4 thrike.dependent variables,salary at entry (time point 1), salary approximately three years later (timepoint 2), andtalary in 1977 ('time point 3), were regressed onto the followingindependent variables: Nucaton,'handicapped veterans,other veteri:7 , Sci"par ,ofage, months of experience, and square of months of Txperfence. The con-trols were of course not incld d in the regression of entry salary.

.To assess the effect of Anority/sex group status on salary, we entered__three dummy variables for.minority/sex (2.-cJo (minority males, minority females,and non-minority females). As the de' -\ c3tegory is non-minority males,the unstandardtzed coefficient for eacn iiimority/sex group should reflect thesalary difference between that group and non-minority males, net of the otherindependent yariables. .This type of .analysis rdOresents, the best jointfitting (or additive) model of /income differences, While an analyiTi5f incomedifferences based on each minority /sex group's statistical earnings structure

Vi

: iD

B-23

. -wouldvpermit an.mexamination of interactions (Taylor, 1979), such /n analysis-.

0 would be ,rather suspect'here, given the small numbers- of minority males and,

..males in/the entry cohorts.

,,, .

t 11. i IP \ . ''' ,

-Pe

-.

4' Poreach cohortthe three'salary variablO ryigld three Jiff regres-sion

. .

analyses- In:the_first of these; entry salary is the dependent variable,v.. ..'-

- *and.the independent variables includ.4he three minority/sex dummy variablesf. .

. and the back'groulid wariWes listed above. In a separate step, we enter fourdumyivariables of occupational. grd0,(Profession41, Administrative, Techaqal,

0.

GleriCal, OgAther--PATCO),to.conirol for the possible effects of job stream(ste SOiler'Man,.0}977 .

. AD.J

.

' , I0,.

; .ft.,.

.

to Fdrithe second multiple regression for each cohort , the' dependent variableis plariapproximately.th.ree years after a cohort entered Federal service, a

4ali...ent.point in.the,aompetitive citil service siobe eNloyees obtain "careero'= . .0!.. status"t,and asso:ated tenure rights after three yearsl.of substantially con-

tinuods service. -.In a second step,'we add another'control variable--entryA salary. .Ih the Eederarservice, entry salary dray be taken as a proxycf9r

i . . ----f,.. ..t% grades and slipice'eate of.promOtion frcr one qradeito the next is regulat

<. ___ *la.L.,_117rOlece'ssa?y to control ttatistitally forntry salary to examine a.ter

..6s4rities due, to internal personnel actions r er than to initial gra eclwiwo

c -40el,of p\ cement! The-PATCO variables are enuredred in a third step of he*..

..§.'regressi5ii,,, gain tC control for job stream,as of that time.

s."',,..'

..... Final*. the third regression for each of the four cohorts. uses 1917

I.o4 salar:y as th dependent variable with the independent variables entered as \..., .

.described ab e for the second set of regression estimates. Since statutoryupper limi4.on Federal salaries greatly reduce the- skewness of salary as cm-

: pared..with individual income, the usual logarithmic transformation of the -

depende:iii variable in such anaA?/ses is not used here. All salary figures areexpres:sed_in 1977 dollars. °

the estimates of salary disparity between non-cAority males and the.

*other three minority/sex groups are .presented in Table 7. Two time points inthe analysis are especially instructive for our purposes.

FirstOthe disparity in entry salary for minority males relative to non-,minority males has decreased over time, as has the salary disparit for non-minority females'. However, the earnings gap at entry has increase for

. minority females, frOm X$1,876 for the 1963-65 cohort to -$2,7 1 for thg-71 cohort. Altbough a control for occupational strewn at entr (PATCO) reducessomeA,at the net earnings gap between'non-minority males and, others, thega0at entry remains ;substantial for both groups of women.

Second, for each cohort tne salary gap three years after entry reveals a. similar trend in deficits accruing to female status. For both groups of women,'1- the salary deficit at this second time point is greater in ttte most recent

cohort-than in the earliest cohort reported. In fact, for minority femilesthe i)attern is a neat, linear one in which the salary disparity increases from

;,,-$2,548 for the'earliest Cohort td -$3,471. for the 1972-73 cohort. For-minority males, however, the salary disparity,_while.alwaysvgativek thesecond, time point, establiShes no clear time trend but chant from -- 03 forthe earliest cohort to -$259 for'the latest.

,A°

B-24 ,

Table 7. Net Economic Detriment to Minority /Sex Status,for Four Entry Cohorts, 1963-1977,'PredictingEntry and Later Salaries.

Cohort TimeEntrants Points

1963-65

(N=831)

1'966-68

(N=1082)

1969-71

(N.679).

1972-73(N=744)

DependentVariablea

(1) Entry Salary

Net of PATCO(2) 1967 Salary

Net of\tntryNet of\PATCO

(3) 1977 Salary

Net of EntryNet. of PATCO,

(1) Entry Salary 0

Net of PATCO(2) 1970 Salary

Net of EntryNet of PATCO

(3), 1977 SalaryNet of EntryNet of PATCO

(1) Entry SalaryNet of PATCO

\(2) 1973 SalaryNet of EntryNet of PATCO

1977 Salary-Net 'of Eary

O. Net of PATCO

(1) Entry SalaryNet of PATCO

(2) 1977 SalaryNet of EntryNet of PATCO

# (3)

Minority/Sex GroupWhite 'Minority MinoritpFemales Males- Females

2,1414**- 1,750***

2,178***- 159

' 18

-4,189***-2,441***2,158***-

2,096*** 4,

-1,459***-2,633***- _457*

161

- 4,138***

- 2,220***

- 4623***

-1,652***- 1,269***-2,650* **

-1,095***692* **

3,647*t*.- 2,218***

-1,454***

I,775***- 1,620***- 2,500***

- 938***711***

-1,054*- 581

- 803183

359-1,478*

- 671

- 29

34

- 165

-1t071*- 459#

- 73

-1,319**-' 813*- 221

- 792- 499

.-1,162*

- 505#

262

7559

298

374

464

259

586*- 309

- 1,876***

1,306***-2,548***9'782 * **

390#

-4,485**w2,971**

- 27705***

- 1,896*w*

-1,246***

'-2,649***

711*5

-4,390***- 2,700***

- 1,791***

- 2,113***

56 * **

3,022***- 1,083***

- 587

_-4,138***

- 2,336t**

- 1,574***

- 2,721***

- 2%463***

- 3,471***

1,140***668*

a: The dependent variables. are standardized to 1977 dollars. The first stepof each regression includes education, D.C., handicapped veterans, otherveterans, age, square,o-rage, and the three minority/sex variables. ATCO and.entry salary are entered on subsequent steps where appropriate. Months ofexperience and squ1are of voths..of experience are enter only in regressions' (2rand f3) for all cohorts. Significance levels are reported as follows:#, p < .1; *,.p < .05; **,,p < .01; and ***, p < .001.1

93

4

B-25

N For all three minority/sex groups, the salary' deficit at the second timepoint i.s generally greater than the salary deficit at entry, suggesting thatinequality is cumulative during the three years from entry to career status.The sole exception to this pattern,is minority males in the 1963-55 cohort.

When controls for entry salary and PATCO are introduced in the regressionpredicting salary three years after entry, the ge eral effect is to reduce thesalary disparity by $500 to 523 C. An exception s minority males in the1972-73 cohort. Even with thes controls, however, thp salary disparities atthe second time point show the ame pattern of intercohort increase as wasobserved without the controls. For example, minority males mho entered inl1963-65 actually experienced a omewhai higher salary than non-minority malesby 1967 once entry salary Ai PATCO are controlled, though the differences arenot statistically significant.

This would suggest that minority males who might have been discriv;natedagainst at entry, were able to hold onto their ititial placement as of fouryears later. Similarly, non-minority females evidence a similar pattern forthe 1963-65 cohort. This same pattern emerges somewhat for the 1966 -6S cohort,and for-females in the 1969-71 cohort. By 1977, however, salaries for the threedisadvantaged mincrity/sex groups in allAphorts showed an increasing distancein salary from non-minority males, even with the initial disadvantage a4 entrystatistically controlled. A

A rather ominous pttern in these data is that abso ute dol ?ar disparitiesseem to be increasing at the three year point from the artiest cohort to thelatest. This suggests that income inequality in the Federa-1 service may begetting worse rather than better, net of occupational placement, for the oldercohorts. For example, in the 1966-68 cohort, the 1970 salaries of minority

41111. and non-minority males were essentially equal, net of entry salary and PATCO.The 1959 -71 Cohort of minority males wa, by 1973, $505 behind non- minoritymales net of the initial salary difference. For the most recent cohort ofminority males, the salary disparity after three years was -5586 controllingfor entry salary. For non-minority females, the salary disparity increase wasfrom a low of -S159 in the 1963-65 cohortat the three year point, to -S938 forthe 1972-73 cohort. Similarly, for minority females the increase was from-$782 to -S1,140. Moreover, this pattern may indicate that in the case ofminority males, especially, salary discrimination has nod from entry placeMentto later in the career cycle. .

Finally, a comparison of the 1977 net salary differences of-the minority/sex groups with non-minority males re-affirms the conclusion that income inequality is cumulative over the career. For,,all minority/sex groups, thelonger one has been in th6 Federal service as of 1977, the greater the 1977salary detriment., For example, non-minori-ty women in the 1963-55dohortexperienced a salary disparity of -$\4,189 with non-minority males by 1977.With PATCO and entry salary controlled, this disparity decreased to -52063."

However, non-minority females.who entered Federal service in 1972-73 and hadtherefogre been in Federal employment only four years instead of 13 yearsexperienced only a - $2,500 difference. Controlling for entry salary and PATCO, .

this disparity dropped to -$711:

In summary, there are four tterns which seem to emerge from thes data.First, income inequality'at entry seems to be declining in the Federal ivil

4

'B-26

0service fortnon-minority women.and minority men. However, minority- womenappear to be worse off in 1973 at entry than in 1963.0s.-a-vis non-minority

.

males. Second, income inequality seems to have.increa$ed by the three year'point in these cohorts, rather than remaihing.static. That even when,differences in entry salary are controlled, the income clap between non-minoritymales and others increases(' after three years of employment. Third, frcm thethree year pointfin each cohort to 1977, income inequality increased across allthree minority/sex groups. Finally, the income inequality of both minority,and non-minority females vis-a-vis non-M,inority males is approximately threetimes as ;reat as that experienced by minority males.

These total differences jn salaryover time substantiate the continuingchallenge to eliminAto pay discrimination. However, it is insttu.Ve toestimate ho1 salary inequality is generated, and one such methodis to co4arethe pay structures themselves over tine. Thit brings .us to 0bjeuive'3:

0(3) What chances have occurredin the pay structures of themirority/sexgroups over time?'

To exeine the pay structure of each minority/sex grobp:the entry salaryeor employees in each minority/ex group masregressed onto education, entryage, entry age squared, D.C. or,field status, disabled veteran, and otherveterans' preference for four cohorts (see also Malkiel and Malkiel, 1973icKluegel, 1978r and Taylor, 1979: for an explanation of this method). Thesevariables most clearly capture the "human capital" of employees at time ofentry into the Federal civil service -, but do not include the effects of occupa-

14.f. tion, ,competitive status, or supervisory pdsition since these latter emOoymentcharacteristics are to a large extent.controlled by the Federal agen:cie them-

Mkselves in the enIploymentprocess.

The results from these analyses are preSented in Table 8. The firstftndingof interest is that e mean-entry Wary of employees,(in 1977 dollars)has decreased over time from e 1963-65 cohort to thea1972-73 cohort. The .

only exception is non- minori females whose mean Salary increased from $8,047to $8,118 over the ten year me span. These data alone suggest that theexternal labor market effects are mediatediv.the,regulations' of the internallabor market, althoijgh the external market has effects where the internal. \

market intersects with it. That is, during the expansionary period of tne'1966 -68 cohort, entry salaries increase4,slightly reflecting igreater demandover supply'of eirployees. Bythe 1972-73, cohort, the dealand f r workers had*fallen beneath supply and entry incomes fell. However,\the 1972-73 entry

,

salary,of minchty males and. both groups of femai,es as, a percent of, their 1963-65 entry` salaries was higher'than tat of white males. These data have notbeen adjasted for changing occupational dittributions or changing job qualifi-cations such as education; Butwthey'do suggest that the Federal givernment mayhe been able to move in the area of affirmative action even while theexternal `labor market was.gettng tiOter.

. .

Looking at the coefficients for education, we can note a decline in-thereturn to one year of S(Oooling at the mean for both groups of males, but,igeneral, an increase for emales. The return to schooing or white males f 11from 57.58 in'the1963\65 cohor to '$550 in the 1972-73, although this declinwas not steady. A simtilar line t? also found for. minority males, from-$814.to $566 That both groups of women's tetuis to education increased ove

,\

©.

/7 Ct.

4

B-27

Table 8. Pay StrUcture of Minority /Sex Groups, for Four Cohorts,1963-1977-,-Pfnicting Entry Salary

.

White'Males'

Indepenctent

Variable,

CQhorts

1963-65 1966-68 1969 -71

,

EaucatronEntry,AgeEntry 'Age2,,/'

D.C.

758***.

1307***-14.97***

2633***

900 * **

,935 ***

-10.i2***

884***196

1788*c-Disabled Vet. -1988# -596, -2021#Other Vet. -1012# -433 89R2 (adjusted) 490 .412 ,.456

300 387 193X Entry Salary $12,843 $12,506 $12,199

iEducation© White 'Entry Age

"Females Entry Age2

Disabled Vet.

438***215**

-3.16**

411

-461Other Vet. y :439

tR2 (adjusted) ,312N '16.9

-X Entry Salary $8:1047

Education .814*Minority Entry Age 124Males Entry Age2 -2.24

: D.C. -2530:--Disabled Vet. -182Other Vet. * 3215R2 (adjusted) .09

N

X Entry Salary

Education 140*Minority Entry Age ,

Females Entry Age2 .,

D.C. -194

553***214*

-2.77#

1042**-598

'1 207

.41ItiAL 252

758,029

1026***222

-2 .

176

.513

105

$8,929

950*** 286-698 596

. 16.26* .° -8.42 ,

674 -301.1396 -71

580 ."' 1553..590 .069

22. 43 23$11,033 $10,163 $9;427

Disabled Vet. -52

Other Vet. 428R2 (adjusted) . .085

N 487( Entry Salary .° $7,522

, k8*** 1008****.

- 281 535#5.6# . -7.92#- 178 -598

-1637-520 231

'.243 .328

100 59:

$7,879 $7,866

1972-.73

550***'

329

-2.41

3433***

-448

-94

.313

254

$11,274.

515***3534

-4.73*45

1433#

.204

237 --

$8,118 '1

566* ,0

1212*

1420

-2410-1924#.253

63

$10,735'

398***352t**

5.21**279

1004

.402 :.

78$6,180

1972-73 EntrySalary as aof 1963-65

87.8.

:100.9

97.3

92.8

(114.1 0

$

IB-28

this period probably reflects the ability of women to enter higher payingmale-dominated occupations in which higher education could be rewarded withhigher salaries.

Given the very small sample sizes of both minority males and females,Conclusions regarding pay differences over time and with white males are

' somewhat tenuous. However, it does appear that the returns to education atentry are generally lower than for white males,'although-the coefficients arenot significantly different. Indeed, by, the 1972-73 cohort, the return to one

,.,year of education at the mean varied from,$393 for minority females to $566 for. minority males. The finding that the returns to education did not vary as

,,,expected by minority/sex group.could'be affected by the construction of thecohorts. Although these data predict the effects of hum,an capital on entrysalary, the persons in this sample consist of those who were still on-board in1977. If the Federal government is a better employer'for minotities and women

*than the private sector, then minorities and women 'will be less likely to leavegovernment service than, for example, white males who might get better returnson their education in the private sector. Only those white males withrelatively high returns to their education would then stay in the Federal

igolrnment.

Also noteworthy is the fact that the 1969-71 cohort shows the highestrate ofreturn to education for women. In fact, it is the only cohort inwhich either group of women had higher returns to education than both groups ofmen. It would be difficult to isolate any one factor as the causal mechanismfor the changeabout, but it seems reasonable to assume that President Johnson's19Q ExatufiVc Order and Executive Order 92-261 (1969) issued by PresidentNixon both acted to sp0r an-,impetus to hire and promote women in the Fedei-alcivil service. The effects of these Executive Orders do not seem exceptionallylong-lived a,the return to education dropped dramatically'for both groups ofwomen by the 1972-73 cohort. .flowever, the returns to education for women inthe 1972-73.entry cohort at least remained greater than the returns to educa-tion for the 1963-65 cohort.-

Although location of employment at the time of entry into the Federalgovernment,would be expected to have an effect on salary, location in the"home office" or D.C. seems to be only consistently positive in its effect onsalary for white males. White males benefit around $2,000 or more for enteringthe Federal civil service in D.C. Other minority/sex groups show no consistentor statisticallysignificant pattern over the cohorts for entering Federalemployment in D.C.

On the. other hand, entry age generally yields about a $300 per yearsalary increase for each year of age, although the exact estimates of returnsto age do vary wisely across the minority/sex grbups and over time. Indeed,both groups of womenseem to have far more stable return's to age than do eithergroup of men. This again may reflect part of the recruiting pattern of malesby age due to the Vietnam War. If 'entry age is increased for males due tomilitary service, and if the relationship between veterans'-preferen6es andsalary is'negative, then,a declining effect of entry age on salary is not sur-prising, provided-that Women enter Federal civil service at younger ages,thanmen.

*1:

R-29

The veterans' preference variables should provide some information toassess the plausibility of the°above arguMent.. However, the°results on theeffect of veterans' preference on entry salary are somewhayrixed. In Contrastto the cross-segtion analysfsA.rformed elsewhere (Taylor, 979), veterans'preference is not strongly rated to salary. Moreover, only 18 of the 28coefficients for disabled veterans and other veterans are negative, and onlyfive of the coefficients approach statistical significance. The lack'ofcorrespondence between the cross-section.results and the cohort analyses pre-sentehere suggests that the negative effects of veterans' preference show uplater in the employment process rather than at entry.

Finally, we should note that the explained varianceaccounted for by onlythese human capital,variables does not appear to be explaining as large a pro-portion of the variance in entry salary as in current salai-y. The reason forthis may also help to explain the different patterns in the returns to educa-tion by sex. If standardized tests haye been used in a greater number of jabhires from 1963 to 1973, it is quite possible that the returns to education at

entry for males would have,declined over this time period since the standard-ized tests are used to both supplement and in some cases, supersede, educationalrequirements. ,This is?not the case for women, however, for even though womenhave begun to move into Male-dominated occtpations, they may do so in su01,small numbers that substituting test score's for educational criteria would havelittle impact. Hence, women are evaluated more on the basis of their education,to be placed into the level of their occupation such as secretary or adminis-trative assistant, while men .may be placed into their entry position by a testscore.

o

In summary, the 1969-71 cohort generally has the largest explained vari-., ance, similar to having the largest effects on returns to education. The uniquecharacteristics of this cohort suggests again the importance of timing in thecareer of an individual employee.

A Review of the Federal ;Ail Service as an Internal Labor Market

The foregoing summary of work funded by the Department of Labor is at oncean examination of the pay determination of Federal civil servants as well as astudy of the extent to which the Federal government operates as a labor marketsimilar to the private sector. That education, age, and occupation'appear tohaveaffecp on salary similar to the private sector is pot surprising, giventhat Federal pay Is set to pay in the private sector. These general findings,as well as others (see Grandjean, 1981), support the argument that in manyrespects the Federal govtrnment is an internal labor market, albeit large andheterogeneous. And although it is a labor market where' seniority seems to be.rewarded there are some indications that career paths are established-early inthe individual's employment history (O'Leary et al., 1978:xii).

There are at leatt two fin)ings which require special Tomment since the'.-Federal government seems to depart froM models of the private sector, and thesedepartures might indicate the extent to which the Federal government does nbtrespond to labor supply and-demand conditions as do private employers. On theother hind, the findings presentedhere'regarding the Federal service may Opvideinsights intotheories of earaings.as we have been able to examine someemployment-related factors not generally available in other studies. Throughoutthis discussion', we will refer to the work of two teams of researchers whohave

0 t93t

k8-30

. -

previously received funding from the Department,of Labur for preliminaryinvestigations into the operations of the Federal government, Couturier et al.,1979, of Northwestern University, and O'Leary, et al., 1978, orOperationSResearch, Incorporated':

One keystone of human 'capital theory is that training, whether in educa-tion,-on-the-job.experience, or vestibule training, enhances a worker's pro-ductivity and plays an important part in wage determination. According toO'Leary (1978:xii), the Federal government significantly outspends itsprivate sector counterpart on training, but little is known about its relationto promotion, mobility, etc. Our findings 'run somewhat counter to.the expecta-:tions of human capital theory. While training is significantly related tosalary for entry employee's, the type of training' receiving the greatest return /is specific rather than general. This finding. may, not be incompatible to humancapital theory, but rather it may suggest the need to incorporate more specific-ally the nature of the fabbr market within which any training is received.4

We know, for example, that Federal personnel policies are -designed tofoster occupational and agency mobility (O'Leary et al., 1978:xtii). This isachieved- by regulations which encourage, through lowered cost and time, the r,internal promotion of employees.. In the Federal government, competitive transferrights mean that employees have more control over mobility than workers in the .

private sector (O'Leary, 1978:13-13). In fact, 1:81 percent of all accession'sare transfers, reassignments, promotions, or reinstatements" (Couturier, et al.,1979:39). This means that those employees who know specific Federal governmepfjoA. are at a decided advantage over employees from outside the-Federal labormarket. Moreover, many programs administered by the Federal governMent-are ofedecidedly political nature and therefore susceptible to changes in politicalddMinistratipns. Keeping employees .with specific training is of ,paramountinterest to a program manager if he or she is to implement and shOw an effect ofa program within a four year span. -

Fi411y, the size of the Federal government -very likely means that there,are subdivisions which compete for.the most talented and well-trained employees(Couturier et al., 1979:8; O'Leary et al., 1978:xiii), Hence, our theories oftraining may need to consider the size of an employer, the number of 404ministea-tive units within a firm, etc., and the personnel yequlations which affectemployee movement within the firm. Given these constraints, we can see thatthe effects of specif4c training and job change ,nay reflect organizational'characteristics rather than simply individual attributes.

We do not think that this finding indicates that the Federal- governmenta labor market too different from private sector, internal labor.markets'"foruseful comparisons. Rather, this finding should be taken as illustrative ofproblems using a human capital perspective without regard ;c) the presence andtype of internJW labor 4rlets in which workers' characteristicS are being ,

evaluated.

,A second finding which may serve to .illustrate the. nature of-the FederArl

bureaucratic labor market is the result of the analysis of minority/s.ex inequal-ity. Since the mid-TKp's, the Federal government has had an active atfirmativeaction progral, In fact, one might argue that the initiation of merit'driterjafor hiring and promotion in- 1883 was the start of practices which would'everit-:-ually lead to non-discrimination Policies; Our findings to income inequality

/13,31

are compatible with those from other studies which suggest that the Federalgovernment rewardojts employees_more 4arly the same than does the privatesector.

An internal labdr market which has a commitment to equal opportunityefforts is more likely to show greater similarity in salary between minorities'and whites, and between men and women, than one which does. not. For example,programs which emphasize enhanced skills for greater employability are morelikely to have *an effect within a labor market where the skills are tied. to ajob._ Such has been the reasoning behind upward mobility programs and training(see Couturier et al., 1979:317; O'Leary et al., 1978:2-16). Training onlybenefits employees and eMployers alike if it provides Something to both which'neither had previously. ThUs, trainjng programs which show the biggest-returns are those'where there is a maxNmum possibility of job advancement.Under affirmative action in the Federal government, upward mobility programs,have been designed to that end. While we do not know whether training receivedby employees is uhder upward mobility programs or not, it 4oes seem more likelythat minorities and women would be in those programs than white males. -Thus,our finding that women and minorities receive,a greater private return to IJTthan do white males may in part be due to the particular programs advanced inthe Federal government. '

This is not to argue that upward mobility programs'are pervasive in thecivil service_or that the.training effect observed here for minorities andwomen is.due directly to such programs. Rather, there is a s-neralized beliefindicated by the presence of EEO efforts, that what minorit es and women needis more and better traininj. Should they receive such training, there is theexpectation that their employability will increase. Hen , training 'effec sare probably _due, to increased skills, a belief 'An the t i ning process its lf,and tht increased contacts and thus greater mobility p ospects'of employees .

who receive training..

9

Conclusions'.

Perhaps as many as one-half to three-Quart rs of all workers in the U.S.are employed'in internal labor markets (Cap-Tow 1954; Doeringer and Piore,1971). with four percent of all workers employed in the Federal civil service.Given the wide variety 4f labor markets in:both product and organization,'.'processes of wage determination are likely'to be affected by many more criteriathan first proposed in studies of job trainfng: This observation does notnecessarily mean that our theories of wage determination are incorrect, only .

that the parameterS'f various labor markets impose constraints on. wage deferml:nation. We argue here that knowing the characteristics of a particular labor ./

market can inform°or understanding of the wage and also refine'ourtheories of labor market behavior.

Our work on the Federal civil service, in part funded by the Department Oflablar,.ts continuing to study the effects of training, minority/4x status, andearnings function of blue-collar versus white-collar workers. We are presentlyat work on a book manuscript which we hope will offer in greater detail inform-.ation on these as well as other areas of interest.

'N v.

k

100

4

N

I. Mee-Collar Workers

APPENDIX I:

Object-it/es and Hypotheses

'6-32

`Hypothesis (1). Years of schooling and amount'of on-the-job training. will be positively associated with salary amoog blue-collar workers.

V,.

*Hypothesis (2). The association in H(1) will be stronger for non-'minority males than for any other minority /sex group.

Objective (1). Does the pay structure (i.e., the determinants of -

earcOngs) of blue-collar workers differ from that o1 white-collar,workers?

Ir. Training ,.

qypothesis (3) . Returr&to training willirbe greater fOr non-minority-.males than for any'other minority/sex group.

.

Hypothe5is-.(4). Individual salary returns to general training willexceed those to specificAraining.

Objective (2). What.is the annual cost of general and specifictrain-ing in dollar outlay for employges who remain in the Federal service

and for thoig who leave? 0

III. Equal EmployMent Opportunity °a.

HypotheOs (5). Among white-collar workers, salary-discrimination byrace and sex his declined from 1967 to 1977.

. Objective (3). What changes have occurred in tile' pay structures of the

minority/sex groups over time?4

I

. -C-2-

Education is measured in approximate,years, from 6 (some elementaryschool) through 22 (post-doctoral 'work,). _Age in years is .computed from yearof birth.

Sdniority refers to 'the number pf months spent in active Federalcivilian employment since. entry. In our cohort analyses, this variable Ye-flects both breaks In service of more than A month and differences in theprecise date of entry, within'the time span defining the cohort. FOr thelongterm employees and in cross-sectional analyses which include pre-1963entrants, exact seniority cannot be Computed from the data available. However,months of Federal service can be inferred with fair accuracy from the "servicecomputation date",maintained on each employee's records for leave accrual ,purposes-. The resulting estimate of seniorityincludes military service formost veterans, but fortunately military duty is not reflected inthe service .

computation date of retired military careerists, for whom the overestimate ofFederal civilian seniority would otherwise be most severe.

Our analyses, also include dummy variables for minority group membership(black, Spanish- surnamed, Puerto Rican, American Indian, Oriental, Aleut, andEskimo) and for:sex-orfor minority/sex group combination.

Blue-collar entry is a dummy variable based on' the eMployePs'first 1;.occupation in, the Federal service. Employees' in a.handful of service occupadons categorized as blue-collar in the -Census_occupational classificationbut paid on a lational salary scalftsuch as the General Schedule are classedas white-collar workers under the OPM schethe (see'Bureau of Policies andStandards., 1978). Since the salary-criterion helps define the boundary, of thewhite-collar Federal labor market, the OPM classification of occupationalcoliar=color is adopted heie.'

two veterans!, preference dummy variablesere included in our analyses,one for disabled-veterans and one for 'all others entitled to.a preference.:

. This approach is chosen to minimize the degree to which the expected positiveeffect of the preference is confounded by a possible negative effect of physicaldisability on socioeconomic achievement.

.

Agencies are grouped into five categorles !lased both on functions and onSimilarities' in patterns of grOwth between 1966 and 1977.. The four resultant,dummy variables.include: .(1) defense-related agencies (Defense and State,plus the SelectiyeService, the National Security Council, and the U.S.Information Agency); 2) the Postal Service; (3) the Veteran's Administration(plus the Soldiers' and Airmen's Home) ;.and (4) a group of social serviceagencies (principally Hp and HUD, plus,the Railroad Retirdment Board, theCommission on Civil Rights, the Equal Employment OppOrtunity Commission, theOffice of-EconOmic Opportunity;, and the Community Services Administration).The comparison category consists of all other departments, agenciet, andcommissions for which OPM maintains personnel records.

,

Finally, employment in the'04.C. metropolitan, area, Civil service "head-quarters,," is-represented by a dummy variable.

102

o

BIBLIOGRAPHY

D -1

Adie, D.K. 0

1977. An Evaluation of Postal Service Wage Rates.. .WashingtOn,D.C.: American Enterprise Institute.

Agency. Compliance and Evaluation Division, Office of Personnel

Management.1979 Federal Civilian Monk Force Statiftics: 'Monthly

Release (Septemb'er). Washington, D.C.: Offite ofPersonnel Management,.

. -

Alexander, A.J. ,

. 1974 "Income, experience, and the structure of internal-labor

markets." Quarterly Journal of.Economics 88 (February):6T-85..

.Alexander, K.L., B.K. Eckland, and C.J. Griffin

0

. 1975 ."The Wisconsin model of socioeconomit'achievement: a

replication." American Journal, of Sociology 8.

(September):324-342.

Almquist, E.M. and J.L. Wehrle-Einhorn.1977, The doubly- disadvantaged: minority women in the labor.

force." Pp.. 63-88 in A.H. Stromberg- and S. Harkess

(editors), WoMen Working: Theories and Facts in

Perspective. Palo Alto, California: Mayfield.

Althauser, R.P., and AA.. Kalleberg.1977 "Occupational and firm labotIlMar.ketos: toward .a specifi-

cation of structural factors'in wage attainment," Paper

read at the annual meeting of the American Sociological

Association.

Alvarez, R. and K.G. Lutterman.1979 Discrimination in Organizations. San Francisco:.

Jossey-Bass.

Asch, SrE., H. Block, and M. Hertzman1938- ' "Studies in the principles of judgments and attitudes

Journal of Psychology 5(April):219-251.

ishenfelter, 0._1978 "EstiMating the effect.of tra ning programs Oil earnings."

4 Review of Economics and Stati cs, 60(February):40.:57.

a

Averitt,, R.T.

1968 The Dual Economy. tiewsYork: Norton.

1fJ3

a.

-

4Babcock, C.

1976 "Basic classification." Presented at Basic. Classification,Bureau .of Training, Thomas Circle, Wdshington, D.C.: (August).

Baron, 0.N and W.T. Bi4lbi. . .

1980 , "Bringing the firms, back in: stratification, segmentation;,. and the organization of work." American Sociological,

Review 45(0Ctober):737-765.1, R

Beck, E.M., P.M. Horan, and C.M. Tolbert, 11.197s ."Stratification.in a dual economy: a sectoral model ofJ

earnings determi ation." American Sociological Review

43(October):704- 20..

. . . .

i

iecker,,G:S.: ,

f 964 Human Capital. New Ycirk: National Bureau of Economic .

Research..

Bendix, R.1949 Higher 'Civil Servants in American Society. University of

Colorado Studies, Series in Sociology, Number 2. Boulder:*

University of'Colorado Press.

1962 Max Weber: An fntellectual Portrait' Garden City, New-York: Anchor.

.,

Benson, C.S., J. Ritzen, and I. Blumenthal-1'974 "Recent perspectives-in the economics of education."

Social Science Quarterly 55(Seritember):244-261'.-

,Berg,'I.

1970 Education and Jobs: :The Great Training Robbery. New

York: Praeger.11

Bergman, B.R. .

1974 'Occupational segregation, wa es, and profit when em-,

ployers dis.criminate by race r sex." Eastern Economic'

Journal 79(AprIl/July):103-1 O.

Bibb, R. and W.H. Form.1977 "The effects of Industrial, occupational, and sex sl6t-

ification on wages in blue-collar markets." Social forces

55(June):974-996.

.Blaloek, H.M.,'Jr.1962 "Occupational,discrimination: some theoretical proposi-

tions." Social Problems 9(Winter) :240-247.

1967 Toward a Theory of Minority-Group Relations: New York:

Wiley.,

1972 %Social Statistics.' New York: McGraw-Hill.-

101

D-3

Blau,, P.M. and 0.D. dunc4n.

'1967 The American Occupational Structure.,,Rew York: Wiley,

and W.R. Scott.1962 Formal Organizations. SanFrancisco: Chandler.

Bluestone, B., W.M. Murphy, and M. Stevenson. gr,1

,',1973 Low Wages and the Working door. Policy-Papers in HumanResources and Industrial Relations 22. Ann,Arbor,Michigan: Ipstitucteu of Labor and Industrial Relations,,

University of Michigan.

Blaug, K.."Y # '

1976 -'"The empirical status of human capital_ theory: a slightly,

jaundiced survey." Journal of Economic Literature.14(September) :1K7-855..

Bdnacich, E.1972 "A theory of ethnic antagonism: the split labor market::

American Sociological Review 37(October):547-559.

1976 "Advanced capitalism and black/w4te,race relations in the

United States: a split labor market interpretation."

American Sociological review 41(Febuary):34-51...

Broom, L. and J.H. 'Smith.1963 g "Bridging occupations." British JOurnal of Sociology

14(December):321-334.

Bureau. of Labor Statistics, U.S: Department of Labor.-1976 Handbook of Labor Statistics 1976. Bulletin 1905.

Washington, D.C.: U.S. Government Printing Office.

Pp. 83-85. '.

1976 Special Labor Force Report 193-. Washington,D.C.:1 Government Printing Office. Pi 12.

1977 Special Labor Force Report 209. Washington,. : U.S.

'Government Printing Office. Pp, 10-11. J1

1977 U.S. Working Women: A Databook. Bulletin 1977.

Washington, D.C.: ,U.S. GovernMent Printing Office. P. 31.

, 1979 employment and Earnings, January. 1979. Volume 26, Number

1. Washington, D.C.: U.S, Govepnment 'Printing Office.

P. T66.

Bureau of Management Services, U.S. Civil Service Commission.1966 Federal Employment of Women. Washington, D.C.: U.S.

Government Printing Office. O. 21. ,

1.05

9

4 .

1

-D-4

. ,,

) ., .,,

Bureau of Manpowei- Informatioh Systems. U.S. Civil ServiceCommission'.-, i,1974 Educatid01 Attainment of General Schedule Employees by . :

. Minority Group and Sex, 4ugust 3i, 1974. Washington, D.t.:,

' U.S. Government Printing Office. P. 20%.

. .,

.1976 Federal Civilian Manpower Statistic's: Monthly Release(November). Washington, D.C.: U.S. Civil ServiceComMission.

.

- ,

40 .1,. . . 1

(

-.Bureau of Personnel Managemnvt InformationSystems, U.S. Civil ServiceCommission. . .

1977-. Equal EmployRnt,Opportunity Statistics, November, 1977..' Washidgton, D.C.: U. Government Rrinting Office.

1977'. Federal Civilian Workforce Statistics: Monthly -Release1

, . September). Washington, D.C..: .U.S% tivil Service

* Commission. .

, T' .

4 -

. -. .

-Bureau of Policies-and Standards, U.S. Civil Service Commission.1970 Position Classification Standards for General Schedule

.._

(GS) Positions. ;Washington, D.C.: U.S. GovernmentPrinting Office.'

1971 Rate of Promotion Study, 1963-1970. Washington; D.C.:

.U.S'. Civil, Service, Commission, Standards Division.

'I,

we"

1978 Census/GS Occupational Crosswalk -(Draft). Washington,

D.C.: U.S. Civil Service Commission, Standards Division,

Bureau;of the Census, U.S. Dep'attment of Commerce.1.975 , Historical Statistics of the United States (Part 2).

Washington, D.C.: U.S. Government Printing Office.

Pp.-1102-1103.

4- -1979 Statistical Abstract of the United States. Washington,

D.C.: U.S. Government Printing Office:.1n

Buredil of Training, U.S. Civil Servite Commission.1977, Report on -Federal Inter-Agency Training. Washington,'

D.C.: U.S. Government Printing Office.

Burstein, P. , ..

P 1979 "Equal employment opportunity legislation and. the income

of women andnonwhites." American Sociological Review

44(June):367-391. -,

Butler, J.S.1976 "Inequality in the military: an examination of the pro-,

motion'time for black and white enlisted men." American

Sociological Review 41(October):807-818. "I-

Byrne, J.J. .

1975 . "Occupational mobility ofworkers,." Monthly LaborReview 98(February).53-59.

Cain, G.G.

D -5

1975 ...'The chhllenge of dual and radical theories ofthe labormarket to orthodox theory.." American Economic Review-65(May):16t22.

14-

1976 "The chal3enge,of segmented labor market theories,to-

. orthpdort,theory: a survey." ...Journal of. Economic

Liter'ature14(Dedember):1215-1257:'

. -

PeP10141. 4

... .1954 The Sociology.of Work. New York:. cGraw-Hill.

,

1

Cayer, N.J.,-.

-

, . 1975 Palic,Personnel. Administration in the United.States.New Ydrk: St. Martin's.

Chinoy, E. .

:

1955 Automobile Workers and the American pream. New,Yonk:'

Doubleday'.

Collins., R.

1975 'Conflict Sociology. -New York: Academic Press.

1979 The Credential Society. New York: Academidjress.

Commission,on Pdlitical Activity of Government Personnel.

1967 A Commission Report: 'Research, Volume 2. Washington, D.C.:

P,S.*Gpvernment Printing.Qffice.

Corazzini, A.J.1972 "Equality of employment opportuhity in the'Federa white-

collar civil service." Journal'of'Human Resources

7(Fall):424-445.

.

, Corcoran, M. '

1978 "Work. experience, work interruption, and 'wages." , In

Five Thousand American Families: PAtterns of Economio

Progress, G. Duncan and J. Mo'rgan-(editors). Ann Arbor,

Michigan: University of Michigan, Institute for Social

Research. Pp. 47-103.

Corspn, J.J.1966

_ I

and R.S. Paul. ,

Men Near the' Top. B41 timore: Johris Hopkins.

" .,

s"...16 .

1:975 "Two methods in search of a substance." A merican ,

'Sociological Review 40(December):691-700.

.1

f

r.

t

Q-6

Couturier, J.J., R.P. Schick, G.S. Goldstein, E.B. McGregor, G.S.,

Maha'ray and C.M. Zashin.

1979 Federal Government Labor Markets: An Approach/to Work-Force Planning. .Final Report to the U:S.,p'epartment of

Labor under contract number DL-20-17-77' -30. Evanston,

Illinois: Northwestern University Graduate Schoo& of

Management.P"

Cutright, P.1974 "The civilian earnings of white and black draftees and

non-veterans." American Sociological Review 40('June):

317-327.

Dalton, M.

1951

7

"Informal factors in career achievement."' AmeriCan

Aournalof Sociology 56(March):407-415.

davis, J.A., Principal Investigator.1975 'Codebook for the Spring 1975 General Social Survey.

Williamstown, Massachusetts: Roger Public Opinion

Research Center.

Davis, K.1948

Davis, R.G.1975

DeVaul,'D.1979.

Human Society. New York: McGraw-Hill.

ancr-G.M. Lewis.Education and Employment. Lexington, Massachusetts:

D.C. Heath.

and W.Gabig:The location of Federal,Civilian Employees:. Policies,

Procedures, and Practices. Report to the U.S. Department

of Housing and Urban Development under contract numberH-4309. Washington, D.C.: Northeast-Midwest Institute.

Doeringer, P.B. and M.J. Piore.4

1971 Internal Labor Markets and Manpower.Analysis.Lexington, Massachus-etts: D.C. Heath.

Downs, A.1967 Inside Bureaucracy. Boston: 'Little, Bl'own.

4k

'Duncan, G. and S. Hoffman.

1978 "Training and earnings." In Five Thousand American

Families: Patterns"5of Economic PrOgress, Voiume'6,

G. Duncan and.J. Morgan (dditorS). Ann-Arbor, Michigan:

- University of Michigan, Institute for Social Research.

105-150.

Duncan, O.D.1975 Introduction to Structural Equation Models. New York:

Academics Press.

1,08

c

D-7'

Eckaus, R.S.1963 "Investment in humah capital: a comment." Journal of

EcOhomy, Volume 71(October):501-504.

111§4 '. "Economic criteria for education and training." Reviewof Economics and Statistics 46(May):181-190:

-Education and Work Group, National Institute of Education. .

1977 Program Plan for Fiscal Years 1977-78. Washington, D.C.:U.S. Department of .'Health, Education and Welfare. Of

.Edwards, R.C.

1975 "The social relations of production in the firm add:labormarket structure." In Labor Market Segmentation, R.Edwards, M. Reich, and D. Gordon (editor's). Leicington,

Massachusetts: D.C. Heath. Pp. 3-26.

Farley, R.1977 "Trends in racial inequalities: Have fbe gains of the

1960s disappeared in the 1970s?" American Sociological.

Review 42(April):189-208.

Featherman, D.L.1976 -' "Coser's...'In search of substance.." The Americall

Sociologist 11(Februavy):2137.

and R.M. Hauser.. 1976b "Sexual inequalities and socioeconomic achievement in

the U.S., 1962-1973." American Sociological Review41(June):462-483.

1976 "Changes in the socioeconomic stratification of the races,

1962-1973." American' Journal of'Socio,logy 82(November):

621 -S51.

1978 Opportunity and Change. New York: Academic Press.

Felkner, B.L..(editor).1978 The U.S. Governments- New York: Bantam.

Fine, S.A.1957 4 -"A reexamination of 'transferability of skills,' Part I."

Monthly Labor Review 80(November): 803-948.

Fitzgerald voHampton.\ 1972 467 F2d 755.

Form, W.H. and J.A. Huber.1976 "Occupational .power. In Handbook of Work, Organization,

and Society, R. Dubin (editor). Chicago: Rand McNally.

- Pp. 751-806.1

109.

et:

D-8

Freedman, Mwith G. MacLachlan.1976 Labor Markets: Segment's and Shelters. Montclair,

New Jersey: Allanheld, Osmun.

, J. Moss and C.W. Taylor.. 1978 "Final report: working panel of consultants. In

F. Pratner,.Occupational Adaptability and TransferableSkills. Information Series Po. 129, The National Centel;forResearch in Vocational Education. Columbus, Ohio:Ohio State University. Pp. 83 -9.1.

Freeman, J.'r

1973 "Origins of the Women's libeefftion movement." AmericanJournal of SOciOlOgY 78(January):792-811.

Gartaganis,1974 , ,"Trends in Federal employment, 1958-72." Monthly Labor

Review 97(October):17-25.

Ginzberg, E.1972 "Toward a theory of occupational choice: a restatement."

Vocational G6idance Quarterly 20(March):169 -175.

1979 Good Jobs, Bad Jobi, No Jobs. Cambridge, Massachusetts:.Harvard University Press.'

Glazer, N.. 1976 Affirmative Discrimination: Ethnic Inequality and POblic

Policy.- New York: Basic Books. Pp. 3 -76.

Glenn, N.D. .

1977 Cohort Analysis. Sage University paper Series 07-005.Beverly Hills: Sage. '

Gordon, D.1972 Theories of Poyerty and Underemployment. Lexington,

Massachusetts:' Heath.

Gordon, R.A.1974 Economic Stability and Growth: The American Record.

New York: Harper and Row.

Grandjean, B.D.1975 "An economic analysis of the Davis-Moore-theory of

stratification." Social ForCes 53(June):543-552.

1981 "History and career in a bureaucratic market." AmericanJournal of Sociology 86 (March):1057-1092.

4Granovetter, M.S.

1974 Getting a Job. Cambridge: Harvard University Prgts.

no

D-§

Grimm, J.W. and R.N. Stern.1974 "Sex roles, and internal labor'market structures: the

.'female' semi-professions.", Social Problems 21(June):690-705.

Grobosky, P.N. and D.H. Rosenbloom.1975 "Racial and ethnic integration in the Federal service."

Social Science Quarterly 56(June):71-84.

Hiller, A.O. and.A. Portes.1973 "Status attainment. processes." -Sociology of Education

'46(Winter):51-91.

Hanock, G.1967 "An economic analysis of earnings and schooling."

Journal of Human Resources 2(Summer):310-329.

.Harrison, B

1972 "Public employment and the theory of the dual economy."In The Political Economy of Public Employment, H.Sheppard, B. Harrison, and W. Spring (editors).

Lexington, Massachusetts: D.C. Heath. Pp. 41-75.

Hauser, R.M., P.J. Dickinson, H.P. Traxi,e, and J.N,. Koffel.

1975 "Structural changes in occupational mobility among men

in the'United States." American 590ological Review40(October):585-598. V''

and D.L. Featherman. :.ite.1977 The PrOcess of Stratification. New York: Academic

Press..

Hellriegel, D. and L. Shoit.'1972. "Equal employment opportunity in the Federal government:

a compArative analysis." Public Administration Review

32(Nov'effber/December):851-858.

Hernes, G.1976

Horan, P.M.1978 "Is status attainment_research atheoretiC

Sociological Review 43(August) :534-54\

"Structural change in social processes." American

Journal of Sociology 82(govember).:513-547. '0

'71 AmeriCan

Hughes, E.C...1937 "Institutional ofritike and thee person."-American Journal, ,

of Sociology 43(November):404-413-, ('5=-...:;;,..'1

1958 Men and Their Wotk. Glencoe., Illi-nbIt: Free Press.

..-**0 ' ,v

s

of

Hunt., S.

1963 Income Determinants for College Graduates and th-e Return

to Educational Investment. Unpublished Ph.D. Disser.tation.

Ya Univeriity.

Interuniversity Consortium for Political and Social Research.

'1978 Guide to Resources and Services. Ann Arbor, Michigan:tnstitute for Survey Research.

Jaffe, A.J..1954

Jencks; C.1980

and R.O. Carleton.Occupational Mobility in the Uni d States: 1930-1960.

New York: King's Criiwn Press.

"Structural versus individual explanations of inequality:',,where_do we go from here?" Contemporary Sociology 9

- (November):762-767. 4

and M..Smith, H. Ackland, M.J. Bane, D. Cohen, H_Gintis

B. ReYns, and S. Michelson.1972 Inequality. New York: Basic Books.

and S. Bartlett, M. Corcoran, J. Crouse, D. Eaglesfield,G. Jackson, K. McClelland, P. Muesser, M.,01neck, J. Schwartz, S. Ward,

and J. William.1979 Who Gets Ahead4 New York: Basic Books.

Job Evaluation and Pay,Review.Task Force, U.S.'Civil Service CommissiOn.

1972 Final' Report of the' Job Evaluation and Pay ReviewTaik .

Force to the United States Civil Service Commission.

Committee Print No. 16, House Committee on Post Office

and Civil SerVite. Washington, D.C.: U.S: Government

. 'Printing Office.

Kalleberg, 6.L. and A.B. Sorensen.1979 "The sociology of labor markets." In Annual Review of

Sociology, Volume 5, A. Inkeles (editor). Palo Alto,

California: Annual Review, Inc. Pp. 351-379.

Kanter, R.M.'1977 Men and Women ofthe,Corporation. New York: Basic

. Books..4h

Kaysen, C.1973

Kelley, J-1973

"New directions for research." Does College Matter?

L. Solmon and P. Taubmah (editors). New York: kademic

Press. Pp. 147-150.

"Causal chain models for the socioeconomic career."American Sociological Review 38(August):481-493.

n2,

o

D-11'

Kerr, C. t

1954 "The ba4anization of labor; markets." Labor MObiTity,

and Economic'Opportunity: Essays. P-.. Webbink (editor).

Cambridge: Massachusetts Institute of Technology.

.Pp. 92-110.

Kilpatrick, F.P., M.C. Cummings, Jr., and M.K. Jennings.,.1964 Source Book of a Study of Occupational Values'and the\Source

of the Federal Service. Washington: Brookings

Institution.

G.C. and Perr=i.1969 . "Social. origins, academic achievetent, and mobility

channel: -sponsored and contest mobility among collegegraduates." Social Forces 48(September) :36 -45.

J.R.

19784 "The causes and costs of racial exclusion from job

authority.' American Sociological Review 43(June):

285-301.

Krislov, S. - .

1967 "Government and equal'employment pppqrtunity." Employment;

Race and Poverty, A.M. Ross and H. Hill (editors). New

York: Harcourt, Brace and World, Inc. Pp. 337 -a64

. 1 .

1967 The Negro in Federal Employment: The, Quest for Equal

. Opportunity. Minneapolis: University of Minnesota

Press.

Lane, A.1975 "The occupational achievement process, 1940-49: a cohort

' analysis." American Sociological, ReView 40(August):

472:482. .

,

Lazear, E.1979 "The narrowing of black-white wage differentials is

illusory." American Economic Review 69(September):

553-564.,

Lepper, M.M. .,,'

,

1974' "A.study of career structures of Federal executives: a

faas on women." Women in Politics, J.S. Jaluette

,(editor). New York\fley, Pp. 109-130.

I

Upset, S.M. and R. Bendix.1952 "Social mobility and occupational career patterns: II.

Social Mobility," American Journal of Sociology 57."

(March) :494-504.

, 0

4.

113

4

.4

D-12

Lloyd, C.B. and B.T. Niemi.1979 The Economics of Sex Differentials. New York: Columbia

University Press.

. Long, J.E..,.(7

,/,..

1976 "'Employment discrimination intfie Federal sector. Journal '

of Human Resources 11(January) :86-97:

Lopreato, J. and L. Alston.

1970 "Ideal types and {the idealization strategy." American

%./ Sociological Review 35(February):88-96.

MacIver, R.M.1947 The Web of Government. New York: Macmillan Company.

Malkiel, B.G. and J.A. Malkiel..1973 "Male-female pay differentialt' in professional employment."

American Economic Review,63(September):693-705,

Martin, N.H. and A.E. Strauss.1959 "Patterns of mobility within industrial orgapizations." In

.. ' Induttrial Man W. Warner and H. Martin (editors).

New York:Jiarper. Pp. 85-101. .

r

McGregor, E.B., Jr.-1974 "Politics and the career mobility of bureaucrats." .

American Political Science Review.68(..January 1:18-26. .

McKinlay, B. , 1.0

1976 Characteristics of Jobs That Aft.Considered Common:Review of Literature an() if 'arch. Columbus:\10bio

State University Center forg oqa,ional4Iducation:, .

Medoff, J.L. and K.G. Abraham.1979 ,"Can productive capacity differentials really explain -,

the earnings differentials associated with demographic

4 characteristics?: the case of experience." .

Mimeo, National Bureau of Economic Research. 4

¶1979 Change n Public Bureaucracies. Cambridge: Cambridge

Uhiversity.Pres,s.

and M.C. Brown.

1977 "The process of bureaucratization." American Joqrnal

.of Sociology 83(September):364-385.

Mill, J.S.'1926 Principals of political Economy. New ,fork :: Longmans,

Green.A

114

A

D-13

Mills, C.W.1959 The Sociological Imagination. London: Oxford University

'Press.

, '0

1962 "On-the-job training: costs, returns, and some implica-

tions." Jo4rnal of Politicl EConomy 70(October):50-7g.°

1970 "The distribution of labor incomes: a survey with special

reference to the human capital approach. 'Journal ofEconomic Literature 8(Spring):1-26..

1974 Schooling, Experience, and Earnings. .New York: National ,

Bureau of Economic Research.

Morison, SrE., H.S. Comrnager, and W.E. Leuchtenbdcg.

1577 A Concise,History of the American Republic. Volume 2:

Since 1865. New York: Oxford.University Press.-

Morton v. Mancari..1974 417 U.S. t35.

Myer;,1975

at, I:"mg

The Coming Col,lapse f the Post'Office.' Englewood Cliffs,

New Jersey: 5rentice-'r

Nachmias,°D. and D.H. Rosenbloom.3980 BUreaucatic Government USA: New 'York; St. Mar'tin's;

Press.

Nigro, L.G.1975

Oaxaca, R.1973

and K.J. Meter."Executive mobility in the federal service:.'a.career

perspective." Public Administration Review 35(t4ay /Jutr')

291 -295.

°.$ -4.1

"Male-female wage differentials' in urban, labor markets.,

International Economic Review 14(October):93-709.

O'Connor, J.19/3 The Fiscal Crisis of the State. New York: St. Ma'rtini's.

Office of Management and $udget.,1980 The BOget of the UnIted States Goiierndent. tiashington,

D.C.: c U.S. Government Printing.Office.d

Office of Personnel Management.1979 General Schedule Series/1970 Census Occupational

Comparison. Federal Personnel Manual Bulletid 40.

511-15. Washington, D.C.: U.S. -Government Pri.nting

\Office.

S

115

4

1,

b

D-14

O'Leary,. J.J., Q.L. McLean, and O.1978 Research Qesign on the Operations of the Federal Gov nment

Labor Market. 'Final Report to the U.S. Department o Labor;

under contract number OL-20-24-77-29.__Sd1veragring,_ 4,

Maryland: Operations Research, Incorporated.

Passett, B.A.1971 Leadership Development fOr Public Service.

Gulf.

,Peabody, R.L. 'and F.E. Rourke.

1965 ,"Public bureaucracies." Handbook of Organizations, J, :March (editor). Chicago: Rand McNally::Pp..§02-W-.1

Perrow,C.r

1979, Complex Organizations. Glenview, Illinois: Scott',

Toresman.

Personnel Research and Development Center, U.S. Civil Service

Commission.1973 Development of a:.Framework for a Factor-Ranking Benchmark

4 System of Job Evaluation. Technical Study TS-7 \3-3.

Washington: U.S.-Civil.Service Commission.

Pfeffer, J.1977 "Toward an examination of stratification1n organizations."

Administrative Science Quarterly 22(pptemberYFW-567.'

OrMe.

1955 Introduction to Labor Economics. New York: . McGraw-Hiff:'

Piore, M.J. `,-

1975 "Notes for a theory of labor market stratification."

. Labor Market Segmentation., R. Edwards, M. ,Reich', and

' . D. Gordon (editors). Lexington, Massachusetts: D.C.

Heath. Pp. 125-150.

Pratzner, F2C7\ .

.,

1978 Occupational Adaptability and Transferable Ski)ls.Infomation SeriesNo. 129, The National Center forReSearch in Vocational Education. Columbus, Ohio: .

. Ohio State University.

and R.W. Stump. .,'

1977 Report on a Project to.Study Occupational Chafge and

. Transferable Skills. Mimed. Paper prepared for an

Intergovernmental Meeting on Vocational Education and, ,

gaining.. .

Prince,. C.E,

197O . "The passing of the arisetocracy: Jefferson's removal of-

the Federalists, 1801.-1805." Journal of American History

57(December):563-575.

4.

A.0 6

4

4

-.,,:., . 0`.... '

,- '-,1-'4-"' t!?..t..-,i,. eo ..,,,

'C ;

Pugh, D.S:,D.J, Hickslonc C:R. Hint%

-C.. riser,1968 "Dimensions of organi , *AA P Administrati.ve :

Science Quarterly 13:0cg.$4,'10'

.,

,-.

',4,...--- r

4 PO

0-15

Reed,-T,L.3972 '"Background or career: whatit. o gets ahead in

the Foreign Service." Paper..tii44 i4,,annual meetingof-the American-Sociowl6gicaliWeititen,

I %iit:A."' -ft t1978 "Organizational change in the Amtilacen Foreign Service,

1925-3965: the.utility of conort*nglysis." .American .

Sociological Review-43 -; -4(Jupe):404-421.

d

Rocke, D.M. and G.W. Downs. ,..

1980 "Time series with episodic disruptions."'Analysing ,

Time Series, 0.D. Andepson (editt4). Amsterdam:. North-

Holland. Pp. 313-326. .,

,

°Rosen, S. . , .

1072 "Learning and experience in the la=bor market." Journal-,

of Human Resources 7(Summer)026---342; ......,-t

.1i..-

Rosenbaum, J.E. %.

.

1979a "Age and college effects on organizational careers: ,alongitbdinal analysis." Pager read at the annual meeting

of the American Sociological Association.

979b "Carder paths and adVancement opportunities." Discrimi-

nation in Organizations, R.,Alvarey and K.G,'Lutterman(editors). San'Francisco: Jossey-Bass. Pp..69-85.

1979c "Organizational career mobi lity: promotion chances in a

corporation during periods of growth and contraction."`American Journal of Sociology 85(JOy):21,48.

Rqsenfeld, R.A.1 1980 "Race and sex differences in career dynamics." American

'Sociological Review 45(August):583-609.

.4;t- "1:

Ross, A.M.1967 "The Negro in the American economy." Employment, Race,

and'Poverty, A.M. Ross and H.-Hill (editors). New York:

Harcourt, Brace., and World. Pp. 3-48.

Ryder, N.B. N 1 r

'1965 "The cohort as a conctpt in the study of social change."American-Sociological Review 30(October):843-861.

Samuelson, P.A.1967 'Economics. New York: McGraw-Hill.

Schneider, P.A.D.1974 The Central Perso nnel Data File (CPDF). Washington, D.C..:

Bureau of Manpower Information Systems,,O.S. Civil Service

Commission.

' 7'

41

Ab.

D-16-

1961 "Investnents in human capital.?-American EcOnomic Review-51 (March):1-17.-

Scism, T.E.1974 "Employee mobility in the Federal service: a description

of some recent data." Public Administration Review 34 .

(May04une))2477254.

Sewell , W.R. andV.P. Shah. 'A.

1967. "Socioeconomic status, intelligence, and the attainmentof higher education." Sociology of Education 40(Wintet):

1 -23.

Shafritz, J.M.1973 Position 'Classification: A Behavioral Analysis for the

Public Service. New York: Praeger.

1975

Sheriff, P.1976 Sociology of Public Bureaucracies 1965-1975. Curi.ent .

Sociology24(December):1-175.

Siegel, P.1,971 Prestige in the American Occupational Structure. Un-.

Public.Personnel Management: The Heritage of Civil

Service'Reform. New York: Praeger.

Sjogren,,D.1977 .

Slocum, W.

' 1966

Smith, J.P.1977

Smith, S.P.N4 1976

published Ph.D. DiFAertation, Urliversity of Chicago.

OccupationallyIATransferable Skills and Characteristics:

Review of Literature and Research. Informatiop Series

No. 105,, The Center for Vocational Education. 'Solumbus,Ohio: Ohio State University.

Occupational Careers: Chicago: Aldine.

and F.R. Welch."-Black-white male wage ratios: 1960-70." American

Economic Review 67(June):323-328.

"Pay differential's between Federal government and private

sector workers." Industrial and Labor Relations Review.

29(January) :179 -197.

1977 Equal Pay in'the Public Sector: Fact or Fantasy.

Princeton, New Jersey: Princeton University Press.

L.C. and P.J': Tlubmari (editors).

1973, Dots College Matter? New York:. Academic Press.

1, 8

. I.

D-17

Sommers,. D. and A. Eck.1977 "Occupational mobility the American labor force."

Monthly tabor Review 10111(January):3-19.

Sorensen, A.B.1974 "A model for occupational 01-eers." Arrierican Journal of

SoOology 80(July):44-57.

1975 ''The structure of intrageheratiOnal mobility.", AmericanSociological Review_40(August):456-471.

Spenner, K4.I. and L.B. Otto:1979 "Emerging issues in the study of careers." Paper read at

the annual meeting of the American Sociological-Association."

SpilermanS.A 1971' "Careers, labor market structure, and socioeconomic'

-)\ achievement." American Journal of Sociology.83-:(Navember):551-593.

Standards Development Center, Office of Persoennel Management.

1980 Comprehensive Analysis for Computer Clerk and AssistantSeries, GS-335. Washington, D.C.: Office.of Personnel

A -Management. . - '....

n.d. Tentative Standards for Crop Insurance'AdministratorSeries, GS-027. Washington, D.C.: 'Office of Personnel

Management.

Stanley, b.T.., AO`'

1964 The Hiper Civil Service., Washington, D.C.: Brookingsp. . , .

Institution. 4c.

.. .

, StiIt

nchcombe, A.L. .

-

, 1959 "'Bureaucratic and craft. Administration of'production,"Administrative Science Guartarly,4(SeptemOer):168-187.

1965 "Social structure and organizations." H4ndbook of

Organizations, J. March (editor). Chicago: Rand

McNally. Pp. 142 -193.

Stone, K.1975 "The origins of job structures. in the steel indtstry."

Labor Market Segmentation, R. Edwards, M. Reich, and D:'Gordon (editors). Lexington, Massachusetts: D.C. Heath.

Pp. 27-84./

Stolzenberg, R.M. , i

1975 "Education, occupation and wage .differences- between whiteand black men." American Journal of Sociology 81(September):299-323:

1975, "Occupations, labor markets, andt-the process of wageattainment." American Sociological Review 40(October):'645-665.

,

. .

8. 11_9

-

D-18

1978 "Bringing'the boss back employerSize, employeeschooling, and socioeconomic achievement." AmericanSociological Review 43(December):813-828.

.,0 Taubman, P.J. ..,1

,1975 -Salines of Inequality in' Earnings: Personal Skills,

Random vents, Preferences Towards Risk and OtherOccupational Characteristics. New York American ElsevierPublishing Company, Inc.

' *

Taylor, F.W. \,

1947 Scientific Management: New York: 'Harper..

Taylor, P.A.- .

. 1979, "Income inequality in the Federal' civilian government."American Sociological Review 44(June):468-479.

19g0 "Institutional job training and 'equal employment opportunity in:

the Federal civil service." Paper presented at 4he Amertcan.

Sdciological Assogiation Meetings, New ,York, New York.

oand N.B. Glenn.

1976 "The utility,of edulotion and attractiveness for females'status attainment tirougbimarriage." American Sociological-Review 41(June):484-4p7:.

and B.D. Grandjean.1979 The 'concept of cohdrt-in the study of inequality." Pjper

prepared for the Research Conference on Phy Equity, .October 23,1979, Washington, D.C. Sponsored by the Center for Women inGovernment (SUNYA), Women Studies Program (GWU), and the Business,and Professional Women's Foundation.

Thurow, Lester C.1969 Poverty and eschmination. Washington, D.C.: The Brookings

Institution.

Treiman,1976 ----JIA-comment on Professor Lewis Coser's presidential

address." The AmeriCan Sociologist 11(February):27-33.

1977 Occupational Prestige in Comparative Perspective, NewYork: AcadeMic Press.

U.S. Civil Service Commissi n.1965 Federal Workfo ce Outlook, Fiscal Years 1966-1969z

Washington, D. .: U.S. Civil Service Commission.,.

1967 Pay Structur f tie Federal,divil Service. Washington,D.C.: U.S. Gdifernment Printing Office.

140

3

4

#

D-19

11976. Personnel Data'Standards. f.PM Supplement 292-1.Washington, D.C.:. i .S. Government Printing Office.

4977 Pay Structure of the Federal Civil Service. Washington,D.C.: U.S. Government Printing Office.

a

U.S. Department of Labor.1972 Handbookfoe Aiplyzing..Jobs. Washington, D.C.: U.S.

Government Printing Office.

1978 National Survey of Professional, Administrative,Tgchnical, and Clerical Pay. Bulletin 2004, Washington?D.C.: U:S. Government Printing Office.

and U.S. Department of 'Health, Education, and Wel978 Employmenand Training Report of the President

Washington, D.C.: U.S. Government Printing Of

U.S. President's Commission on the'Status of Women.1963 Report of the Committee on Federal Employment.

D.C.: U.S. Government Printing Office.,.

Van Horn, C.W.G. f

1972 The Hay Guide Chart-Profile.Method:" Handbook of Wage .

, and Salary Adm-inisti-ation, M. Rock (editor). New y,ork:

McGraw-Hill. -Pp. 86-97.

Van Riper, P.P.1-1958 bliistory of the United States Civil° Service. White Plains,

New York: Row Peterson.

Warner, 'P.P. VanRiper, N.H. Martin, and O.F. Collins.196 The American Federal Executive. New Haven: Yale .

University Press: ,

fare.

fice.

Wathington,

Weber, M.1568 Economy and Society. Edited by G. Roth and C. Wittich.

New York: Beamjnstetl.

e,

'Whitg, H.C..1970 Chains of .Opportunity. Cambridge, Massachusetts:41arvahl

University Press,;',-,

-

Wilensky, H.L.. -

ft

1960 "Work, careers, and social in tegration." -InternationalSocial Science Journal' 12(Fall):543-560:v*

1961 ' rderly career and social participation: the impact ofark history on social integration in the middle class."

American Sociological Review 26(August):521-539,

Williams, R.1976 Keywords: A Vocabulary of Culture and Society. New

York: Oxford University Press.

121

D-20

1011, P. and R. Jones.1975 "Bureaucratic defense in depthg.".4Wat6rgate and the

American Political Process, R. Pynn(editor). New York:Praeger. Pp. 216-224.

Wright, E.O.1978 "Race, class,.and income inequality." American Journal

of Sociology 83(May):1368-1397:

X22

.